1 \documentclass[USenglish,11pt]{ifimaster}
3 \usepackage[utf8]{inputenc}
4 \usepackage[T1]{fontenc,url}
5 \usepackage{lmodern} % using Latin Modern to be able to use bold typewriter font
12 \usepackage{tikz-qtree}
13 \usetikzlibrary{shapes,snakes,trees,arrows,shadows,positioning,calc}
14 \usepackage{babel,textcomp,csquotes,ifimasterforside}
17 \usepackage[hidelinks]{hyperref}
19 \usepackage[xindy]{glossaries}
21 \usepackage[style=alphabetic,backend=biber,doi=false,isbn=false]{biblatex}
23 \usepackage{mathtools}
25 % use 'disable' before printing:
26 \usepackage[]{todonotes}
34 \def\mintedframesep{11pt}
36 \usepackage{perpage} %the perpage package
37 \MakePerPage{footnote} %the perpage package command
39 \theoremstyle{definition}
40 \newtheorem*{wordDef}{Definition}
41 \newtheorem*{theorem}{Theorem}
43 \graphicspath{ {./figures/} }
45 \newcommand{\citing}[1]{~\cite{#1}}
46 %\newcommand{\myref}[1]{\cref{#1} on \cpageref{#1}}
47 \newcommand{\myref}[1]{\vref{#1}}
48 \newcommand{\Myref}[1]{\Vref{#1}}
49 \newcommand{\mysimpleref}[1]{\cref{#1}}
50 \newcommand{\Mysimpleref}[1]{\Cref{#1}}
52 %\newcommand{\glossref}[1]{\textsuperscript{(\glsrefentry{#1})}}
53 %\newcommand{\gloss}[1]{\gls{#1}\glossref{#1}}
54 %\newcommand{\glosspl}[1]{\glspl{#1}\glossref{#1}}
55 \newcommand{\gloss}[1]{\gls{#1}}
56 \newcommand{\glosspl}[1]{\glspl{#1}}
58 \newcommand{\definition}[1]{\begin{wordDef}#1\end{wordDef}}
59 \newcommand{\see}[1]{(see \myref{#1})}
60 \newcommand{\explanation}[3]{\noindent\textbf{\textit{#1}}\\*\emph{When:}
61 #2\\*\emph{How:} #3\\*[-7px]}
63 %\newcommand{\type}[1]{\lstinline{#1}}
64 \newcommand{\code}[1]{\texttt{\textbf{#1}}}
65 \newcommand{\type}[1]{\code{#1}}
66 \newcommand{\typeref}[1]{\footnote{\type{#1}}}
67 \newcommand{\typewithref}[2]{\type{#2}\typeref{#1.#2}}
68 \newcommand{\method}[1]{\type{#1}}
69 \newcommand{\methodref}[2]{\footnote{\type{#1}\method{\##2()}}}
70 \newcommand{\methodwithref}[2]{\method{#2}\footnote{\type{#1}\method{\##2()}}}
71 \newcommand{\var}[1]{\type{#1}}
73 \newcommand{\name}[1]{#1}
74 \newcommand{\tit}[1]{\emph{#1}}
75 \newcommand{\refa}[1]{\emph{#1}}
76 \newcommand{\pattern}[1]{\emph{#1}}
77 \newcommand{\metr}[1]{\emph{#1}}
78 \newcommand{\ExtractMethod}{\refa{Extract Method}\xspace}
79 \newcommand{\MoveMethod}{\refa{Move Method}\xspace}
80 \newcommand{\ExtractAndMoveMethod}{\refa{Extract and Move Method}\xspace}
82 \newcommand{\m}[1]{$#1$}
84 \newcommand\todoin[2][]{\todo[inline, caption={#2}, #1]{
85 \begin{minipage}{\textwidth-4pt}#2\end{minipage}}}
87 \title{Automated Composition of Refactorings}
88 \subtitle{Implementing and evaluating a search-based Extract and Move Method
90 \author{Erlend Kristiansen}
93 \newglossaryentry{profiling}
96 description={is to run a computer program through a profiler/with a profiler
99 \newglossaryentry{profiler}
102 description={A profiler is a program for analyzing performance within an
103 application. It is used to analyze memory consumption, processing time and
104 frequency of procedure calls and such}
106 \newglossaryentry{xUnit}
108 name={xUnit framework},
109 description={An xUnit framework is a framework for writing unit tests for a
110 computer program. It follows the patterns known from the JUnit framework for
111 Java\citing{fowlerXunit}
113 plural={xUnit frameworks}
115 \newglossaryentry{softwareObfuscation}
117 name={software obfuscation},
118 description={makes source code harder to read and analyze, while preserving
121 \newglossaryentry{extractClass}
123 name=\refa{Extract Class},
124 description={The \refa{Extract Class} refactoring works by creating a class,
125 for then to move members from another class to that class and access them from
126 the old class via a reference to the new class}
128 \newglossaryentry{designPattern}
130 name={design pattern},
131 description={A design pattern is a named abstraction that is meant to solve a
132 general design problem. It describes the key aspects of a common problem and
133 identifies its participators and how they collaborate},
134 plural={design patterns}
136 \newglossaryentry{enclosingClass}
138 name={enclosing class},
139 description={An enclosing class is the class that surrounds any specific piece
140 of code that is written in the inner scope of this class},
142 \newglossaryentry{mementoPattern}
144 name={memento pattern},
145 description={The memento pattern is a software design pattern that is used to
146 capture an object's internal state so that it can be restored to this state
147 later\citing{designPatterns}},
149 %\newglossaryentry{extractMethod}
151 % name=\refa{Extract Method},
152 % description={The \refa{Extract Method} refactoring is used to extract a
153 %fragment of code from its context and into a new method. A call to the new
154 %method is inlined where the fragment was before. It is used to break code into
155 %logical units, with names that explain their purpose}
157 %\newglossaryentry{moveMethod}
159 % name=\refa{Move Method},
160 % description={The \refa{Move Method} refactoring is used to move a method from
161 % one class to another. This is useful if the method is using more features of
162 % another class than of the class which it is currently defined. Then all calls
163 % to this method must be updated, or the method must be copied, with the old
164 %method delegating to the new method}
167 \bibliography{bibliography/master-thesis-erlenkr-bibliography}
168 \DefineBibliographyStrings{english}{%
169 bibliography = {References},
171 \newbibmacro{string+doi}[1]{%
172 \iffieldundef{doi}{#1}{\href{http://dx.doi.org/\thefield{doi}}{#1}}}
173 \DeclareFieldFormat{title}{\usebibmacro{string+doi}{\mkbibemph{#1}}}
174 \DeclareFieldFormat[article]{title}{\usebibmacro{string+doi}{\mkbibquote{#1}}}
176 % UML comment in TikZ:
177 % ref: https://tex.stackexchange.com/questions/103688/folded-paper-shape-tikz
179 \pgfdeclareshape{umlcomment}{
180 \inheritsavedanchors[from=rectangle] % this is nearly a rectangle
181 \inheritanchorborder[from=rectangle]
182 \inheritanchor[from=rectangle]{center}
183 \inheritanchor[from=rectangle]{north}
184 \inheritanchor[from=rectangle]{south}
185 \inheritanchor[from=rectangle]{west}
186 \inheritanchor[from=rectangle]{east}
187 % ... and possibly more
188 \backgroundpath{% this is new
189 % store lower right in xa/ya and upper right in xb/yb
190 \southwest \pgf@xa=\pgf@x \pgf@ya=\pgf@y
191 \northeast \pgf@xb=\pgf@x \pgf@yb=\pgf@y
192 % compute corner of ‘‘flipped page’’
193 \pgf@xc=\pgf@xb \advance\pgf@xc by-10pt % this should be a parameter
194 \pgf@yc=\pgf@yb \advance\pgf@yc by-10pt
195 % construct main path
196 \pgfpathmoveto{\pgfpoint{\pgf@xa}{\pgf@ya}}
197 \pgfpathlineto{\pgfpoint{\pgf@xa}{\pgf@yb}}
198 \pgfpathlineto{\pgfpoint{\pgf@xc}{\pgf@yb}}
199 \pgfpathlineto{\pgfpoint{\pgf@xb}{\pgf@yc}}
200 \pgfpathlineto{\pgfpoint{\pgf@xb}{\pgf@ya}}
203 \pgfpathmoveto{\pgfpoint{\pgf@xc}{\pgf@yb}}
204 \pgfpathlineto{\pgfpoint{\pgf@xc}{\pgf@yc}}
205 \pgfpathlineto{\pgfpoint{\pgf@xb}{\pgf@yc}}
206 \pgfpathlineto{\pgfpoint{\pgf@xc}{\pgf@yc}}
211 \tikzstyle{comment}=[%
224 %\interfootnotelinepenalty=10000
226 % Space between table rows
227 \renewcommand{\arraystretch}{1.3}
229 \newcommand{\spancols}[2]{\multicolumn{#1}{@{}l@{}}{#2}}
231 \newcolumntype{L}[1]{>{\hsize=#1\hsize\raggedright\arraybackslash}X}%
232 \newcolumntype{R}[1]{>{\hsize=#1\hsize\raggedleft\arraybackslash}X}%
234 \setcounter{tocdepth}{1}
237 %\pagenumbering{arabic}
242 %\setcounter{page}{3}
245 \todoin{\textbf{Remove all todos (including list) before delivery/printing!!!
246 Can be done by removing ``draft'' from documentclass.}}
247 \todoin{Write abstract}
255 %\setcounter{page}{13}
257 \chapter{Introduction}
259 \section{Motivation and structure}
261 For large software projects, complex program source code is an issue. It impacts
262 the cost of maintenance in a negative way. It often stalls the implementation of
263 new functionality and other program changes. The code may be difficult to
264 understand, the changes may introduce new bugs that are hard to find and its
265 complexity can simply keep people from doing code changes in fear of breaking
266 some dependent piece of code. All these problems are related, and often lead to
267 a vicious circle that slowly degrades the overall quality of a project.
269 More specifically, and in an object-oriented context, a class may depend on a
270 number of other classes. Sometimes these intimate relationships are appropriate,
271 and sometimes they are not. Inappropriate \emph{coupling} between classes can
272 make it difficult to know whether or not a change that is aimed at fixing a
273 specific problem also alters the behavior of another part of a program.
275 One of the tools that are used to fight complexity and coupling in program
276 source code is \emph{refactoring}. The intention for this master's thesis is
277 therefore to create an automated composite refactoring that reduces coupling
278 between classes. The refactoring shall be able to operate automatically in all
279 phases of a refactoring, from performing analysis to executing changes. It is
280 also a requirement that it should be able to process large quantities of source
281 code in a reasonable amount of time.
283 The current chapter proceeds in \mysimpleref{sec:refactoring} by describing what
284 refactoring is. Then the project is presented in \mysimpleref{sec:project},
285 before the chapter is concluded with a brief discussion of related work in
286 \mysimpleref{sec:relatedWork}.
288 \Mysimpleref{ch:extractAndMoveMethod} shows the workings of our refactoring
289 together with a simple example illustrating this.
291 \todoin{Structure. Write later\ldots}
294 \section{What is refactoring?}\label{sec:refactoring}
296 This question is best answered by first defining the concept of a
297 \emph{refactoring}, what it is to \emph{refactor}, and then discuss what aspects
298 of programming make people want to refactor their code.
300 \subsection{Defining refactoring}
301 Martin Fowler, in his classic book on refactoring\citing{refactoring}, defines a
302 refactoring like this:
305 \emph{Refactoring} (noun): a change made to the internal
306 structure\footnote{The structure observable by the programmer.} of software to
307 make it easier to understand and cheaper to modify without changing its
308 observable behavior.~\cite[p.~53]{refactoring}
311 \noindent This definition assigns additional meaning to the word
312 \emph{refactoring}, beyond the composition of the prefix \emph{re-}, usually
313 meaning something like ``again'' or ``anew'', and the word \emph{factoring},
314 which can mean to isolate the \emph{factors} of something. Here a \emph{factor}
315 would be close to the mathematical definition of something that divides a
316 quantity, without leaving a remainder. Fowler is mixing the \emph{motivation}
317 behind refactoring into his definition. Instead it could be more refined, formed
318 to only consider the \emph{mechanical} and \emph{behavioral} aspects of
319 refactoring. That is to factor the program again, putting it together in a
320 different way than before, while preserving the behavior of the program. An
321 alternative definition could then be:
323 \definition{A \emph{refactoring} is a transformation
324 done to a program without altering its external behavior.}
326 From this we can conclude that a refactoring primarily changes how the
327 \emph{code} of a program is perceived by the \emph{programmer}, and not the
328 \emph{behavior} experienced by any user of the program. Although the logical
329 meaning is preserved, such changes could potentially alter the program's
330 behavior when it comes to performance gain or -penalties. So any logic depending
331 on the performance of a program could make the program behave differently after
334 In the extreme case one could argue that \gloss{softwareObfuscation} is
335 refactoring. It is often used to protect proprietary software. It restrains
336 uninvited viewers, so they have a hard time analyzing code that they are not
337 supposed to know how works. This could be a problem when using a language that
338 is possible to decompile, such as Java.
340 Obfuscation could be done composing many, more or less randomly chosen,
341 refactorings. Then the question arises whether it can be called a
342 \emph{composite refactoring} or not \see{compositeRefactorings}? The answer is
343 not obvious. First, there is no way to describe the mechanics of software
344 obfuscation, because there are infinitely many ways to do that. Second,
345 obfuscation can be thought of as \emph{one operation}: Either the code is
346 obfuscated, or it is not. Third, it makes no sense to call software obfuscation
347 \emph{a refactoring}, since it holds different meaning to different people.
349 This last point is important, since one of the motivations behind defining
350 different refactorings, is to establish a \emph{vocabulary} for software
351 professionals to use when reasoning about and discussing programs, similar to
352 the motivation behind \glosspl{designPattern}\citing{designPatterns}.
354 So for describing \emph{software obfuscation}, it might be more appropriate to
355 define what you do when performing it rather than precisely defining its
356 mechanics in terms of other refactorings.
359 \subsection{The etymology of 'refactoring'}
360 It is a little difficult to pinpoint the exact origin of the word
361 ``refactoring'', as it seems to have evolved as part of a colloquial
362 terminology, more than a scientific term. There is no authoritative source for a
363 formal definition of it.
365 According to Martin Fowler\citing{etymology-refactoring}, there may also be more
366 than one origin of the word. The most well-known source, when it comes to the
367 origin of \emph{refactoring}, is the
368 Smalltalk\footnote{\label{footNote}Programming language} community and their
369 infamous \name{Refactoring
370 Browser}\footnote{\url{http://st-www.cs.illinois.edu/users/brant/Refactory/RefactoringBrowser.html}}
371 described in the article \tit{A Refactoring Tool for
372 Smalltalk}\citing{refactoringBrowser1997}, published in 1997.
373 Allegedly\citing{etymology-refactoring}, the metaphor of factoring programs was
374 also present in the Forth\textsuperscript{\ref{footNote}} community, and the
375 word ``refactoring'' is mentioned in a book by Leo Brodie, called \tit{Thinking
376 Forth}\citing{brodie2004}, first published in 1984\footnote{\tit{Thinking Forth}
377 was first published in 1984 by the \name{Forth Interest Group}. Then it was
378 reprinted in 1994 with minor typographical corrections, before it was
379 transcribed into an electronic edition typeset in \LaTeX\ and published under a
380 Creative Commons license in
381 2004. The edition cited here is the 2004 edition, but the content should
382 essentially be as in 1984.}. The exact word is only printed one
383 place~\cite[p.~232]{brodie2004}, but the term \emph{factoring} is prominent in
384 the book, which also contains a whole chapter dedicated to (re)factoring, and
385 how to keep the (Forth) code clean and maintainable.
388 \ldots good factoring technique is perhaps the most important skill for a
389 Forth programmer.~\cite[p.~172]{brodie2004}
392 \noindent Brodie also express what \emph{factoring} means to him:
395 Factoring means organizing code into useful fragments. To make a fragment
396 useful, you often must separate reusable parts from non-reusable parts. The
397 reusable parts become new definitions. The non-reusable parts become arguments
398 or parameters to the definitions.~\cite[p.~172]{brodie2004}
401 Fowler claims that the usage of the word \emph{refactoring} did not pass between
402 the \name{Forth} and \name{Smalltalk} communities, but that it emerged
403 independently in each of the communities.
405 \subsection{Reasons for refactoring}
406 There are many reasons why people want to refactor their programs. They can for
407 instance do it to remove duplication, break up long methods or to introduce
408 design patterns into their software systems. The shared trait for all these is
409 that peoples' intentions are to make their programs \emph{better}, in some
410 sense. But what aspects of their programs are becoming improved?
412 As just mentioned, people often refactor to get rid of duplication. They are
413 moving identical or similar code into methods, and are pushing methods up or
414 down in their class hierarchies. They are making template methods for
415 overlapping algorithms/functionality, and so on. It is all about gathering what
416 belongs together and putting it all in one place. The resulting code is then
417 easier to maintain. When removing the implicit coupling\footnote{When
418 duplicating code, the duplicate pieces of code might not be coupled, apart
419 from representing the same functionality. So if this functionality is going to
420 change, it might need to change in more than one place, thus creating an
421 implicit coupling between multiple pieces of code.} between code snippets, the
422 location of a bug is limited to only one place, and new functionality need only
423 to be added to this one place, instead of a number of places people might not
426 A problem you often encounter when programming, is that a program contains a lot
427 of long and hard-to-grasp methods. It can then help to break the methods into
428 smaller ones, using the \ExtractMethod refactoring\citing{refactoring}. Then
429 you may discover something about a program that you were not aware of before;
430 revealing bugs you did not know about or could not find due to the complex
431 structure of your program. Making the methods smaller and giving good names to
432 the new ones clarifies the algorithms and enhances the \emph{understandability}
433 of the program \see{magic_number_seven}. This makes refactoring an excellent
434 method for exploring unknown program code, or code that you had forgotten that
437 Most primitive refactorings are simple, and usually involves moving code
438 around\citing{kerievsky2005}. The motivation behind them may first be revealed
439 when they are combined into larger --- higher level --- refactorings, called
440 \emph{composite refactorings} \see{compositeRefactorings}. Often the goal of
441 such a series of refactorings is a design pattern. Thus the design can
442 \emph{evolve} throughout the lifetime of a program, as opposed to designing
443 up-front. It is all about being structured and taking small steps to improve a
446 Many software design pattern are aimed at lowering the coupling between
447 different classes and different layers of logic. One of the most famous is
448 perhaps the \pattern{Model-View-Controller}\citing{designPatterns} pattern. It
449 is aimed at lowering the coupling between the user interface, the business logic
450 and the data representation of a program. This also has the added benefit that
451 the business logic could much easier be the target of automated tests, thus
452 increasing the productivity in the software development process.
454 Another effect of refactoring is that with the increased separation of concerns
455 coming out of many refactorings, the \emph{performance} can be improved. When
456 profiling programs, the problematic parts are narrowed down to smaller parts of
457 the code, which are easier to tune, and optimization can be performed only where
458 needed and in a more effective way\citing{refactoring}.
460 Last, but not least, and this should probably be the best reason to refactor, is
461 to refactor to \emph{facilitate a program change}. If one has managed to keep
462 one's code clean and tidy, and the code is not bloated with design patterns that
463 are not ever going to be needed, then some refactoring might be needed to
464 introduce a design pattern that is appropriate for the change that is going to
467 Refactoring program code --- with a goal in mind --- can give the code itself
468 more value. That is in the form of robustness to bugs, understandability and
469 maintainability. Having robust code is an obvious advantage, but
470 understandability and maintainability are both very important aspects of
471 software development. By incorporating refactoring in the development process,
472 bugs are found faster, new functionality is added more easily and code is easier
473 to understand by the next person exposed to it, which might as well be the
474 person who wrote it. The consequence of this, is that refactoring can increase
475 the average productivity of the development process, and thus also add to the
476 monetary value of a business in the long run. The perspective on productivity
477 and money should also be able to open the eyes of the many nearsighted managers
478 that seldom see beyond the next milestone.
480 \subsection{The magical number seven}\label{magic_number_seven}
481 The article \tit{The magical number seven, plus or minus two: some limits on our
482 capacity for processing information}\citing{miller1956} by George A. Miller,
483 was published in the journal \name{Psychological Review} in 1956. It presents
484 evidence that support that the capacity of the number of objects a human being
485 can hold in its working memory is roughly seven, plus or minus two objects. This
486 number varies a bit depending on the nature and complexity of the objects, but
487 is according to Miller ``\ldots never changing so much as to be
490 Miller's article culminates in the section called \emph{Recoding}, a term he
491 borrows from communication theory. The central result in this section is that by
492 recoding information, the capacity of the amount of information that a human can
493 process at a time is increased. By \emph{recoding}, Miller means to group
494 objects together in chunks, and give each chunk a new name that it can be
498 \ldots recoding is an extremely powerful weapon for increasing the amount of
499 information that we can deal with.~\cite[p.~95]{miller1956}
502 By organizing objects into patterns of ever growing depth, one can memorize and
503 process a much larger amount of data than if it were to be represented as its
504 basic pieces. This grouping and renaming is analogous to how many refactorings
505 work, by grouping pieces of code and give them a new name. Examples are the
506 fundamental \ExtractMethod and \refa{Extract Class}
507 refactorings\citing{refactoring}.
509 An example from the article addresses the problem of memorizing a sequence of
510 binary digits. The example presented here is a slightly modified version of the
511 one presented in the original article\citing{miller1956}, but it preserves the
512 essence of it. Let us say we have the following sequence of
513 16 binary digits: ``1010001001110011''. Most of us will have a hard time
514 memorizing this sequence by only reading it once or twice. Imagine if we instead
515 translate it to this sequence: ``A273''. If you have a background from computer
516 science, it will be obvious that the latter sequence is the first sequence
517 recoded to be represented by digits in base 16. Most people should be able to
518 memorize this last sequence by only looking at it once.
520 Another result from the Miller article is that when the amount of information a
521 human must interpret increases, it is crucial that the translation from one code
522 to another must be almost automatic for the subject to be able to remember the
523 translation, before \heshe is presented with new information to recode. Thus
524 learning and understanding how to best organize certain kinds of data is
525 essential to efficiently handle that kind of data in the future. This is much
526 like when humans learn to read. First they must learn how to recognize letters.
527 Then they can learn distinct words, and later read sequences of words that form
528 whole sentences. Eventually, most of them will be able to read whole books and
529 briefly retell the important parts of its content. This suggests that the use of
530 design patterns is a good idea when reasoning about computer programs. With
531 extensive use of design patterns when creating complex program structures, one
532 does not always have to read whole classes of code to comprehend how they
533 function, it may be sufficient to only see the name of a class to almost fully
534 understand its responsibilities.
537 Our language is tremendously useful for repackaging material into a few chunks
538 rich in information.~\cite[p.~95]{miller1956}
541 Without further evidence, these results at least indicate that refactoring
542 source code into smaller units with higher cohesion and, when needed,
543 introducing appropriate design patterns, should aid in the cause of creating
544 computer programs that are easier to maintain and have code that is easier (and
547 \subsection{Notable contributions to the refactoring literature}
550 \item[1992] William F. Opdyke submits his doctoral dissertation called
551 \tit{Refactoring Object-Oriented Frameworks}\citing{opdyke1992}. This work
552 defines a set of refactorings that are behavior-preserving given that their
553 preconditions are met. The dissertation is focused on the automation of
555 \item[1999] Martin Fowler et al.: \tit{Refactoring: Improving the Design of
556 Existing Code}\citing{refactoring}. This is maybe the most influential text
557 on refactoring. It bares similarities with Opdykes thesis\citing{opdyke1992}
558 in the way that it provides a catalog of refactorings. But Fowler's book is
559 more about the craft of refactoring, as he focuses on establishing a
560 vocabulary for refactoring, together with the mechanics of different
561 refactorings and when to perform them. His methodology is also founded on
562 the principles of test-driven development.
563 \item[2005] Joshua Kerievsky: \tit{Refactoring to
564 Patterns}\citing{kerievsky2005}. This book is heavily influenced by Fowler's
565 \tit{Refactoring}\citing{refactoring} and the ``Gang of Four'' \tit{Design
566 Patterns}\citing{designPatterns}. It is building on the refactoring
567 catalogue from Fowler's book, but is trying to bridge the gap between
568 \emph{refactoring} and \emph{design patterns} by providing a series of
569 higher-level composite refactorings, that makes code evolve toward or away
570 from certain design patterns. The book is trying to build up the reader's
571 intuition around \emph{why} one would want to use a particular design
572 pattern, and not just \emph{how}. The book is encouraging evolutionary
573 design \see{relationToDesignPatterns}.
576 \subsection{Tool support (for Java)}\label{toolSupport}
577 This section will briefly compare the refactoring support of the three IDEs
578 \name{Eclipse}\footnote{\url{http://www.eclipse.org/}}, \name{IntelliJ
579 IDEA}\footnote{The IDE under comparison is the \name{Community Edition},
580 \url{http://www.jetbrains.com/idea/}} and
581 \name{NetBeans}\footnote{\url{https://netbeans.org/}}. These are the most
582 popular Java IDEs\citing{javaReport2011}.
584 All three IDEs provide support for the most useful refactorings, like the
585 different extract, move and rename refactorings. In fact, Java-targeted IDEs are
586 known for their good refactoring support, so this did not appear as a big
589 The IDEs seem to have excellent support for the \ExtractMethod refactoring, so
590 at least they have all passed the first ``refactoring
591 rubicon''\citing{fowlerRubicon2001,secondRubicon2012}.
593 Regarding the \MoveMethod refactoring, the \name{Eclipse} and \name{IntelliJ}
594 IDEs do the job in very similar manners. In most situations they both do a
595 satisfying job by producing the expected outcome. But they do nothing to check
596 that the result does not break the semantics of the program
597 \see{sec:correctness}.
598 The \name{NetBeans} IDE implements this refactoring in a somewhat
599 unsophisticated way. For starters, the refactoring's default destination for the
600 move, is the same class as the method already resides in, although it refuses to
601 perform the refactoring if chosen. But the worst part is, that if moving the
602 method \method{f} of the class \type{C} to the class \type{X}, it will break the
603 code. The result is shown in \myref{lst:moveMethod_NetBeans}.
607 \begin{minted}[samepage]{java}
620 \begin{minted}[samepage]{java}
630 \caption{Moving method \method{f} from \type{C} to \type{X}.}
631 \label{lst:moveMethod_NetBeans}
634 \name{NetBeans} will try to create code that call the methods \method{m} and \method{n}
635 of \type{X} by accessing them through \var{c.x}, where \var{c} is a parameter of
636 type \type{C} that is added the method \method{f} when it is moved. (This is
637 seldom the desired outcome of this refactoring, but ironically, this ``feature''
638 keeps \name{NetBeans} from breaking the code in the example from
639 \myref{sec:correctness}.) If \var{c.x} for some reason is inaccessible to
640 \type{X}, as in this case, the refactoring breaks the code, and it will not
641 compile. \name{NetBeans} presents a preview of the refactoring outcome, but the
642 preview does not catch it if the IDE is about break the program.
644 The IDEs under investigation seem to have fairly good support for primitive
645 refactorings, but what about more complex ones, such as
646 \gloss{extractClass}\citing{refactoring}? \name{IntelliJ} handles this in a
647 fairly good manner, although, in the case of private methods, it leaves unused
648 methods behind. These are methods that delegate to a field with the type of the
649 new class, but are not used anywhere. \name{Eclipse} has added its own quirk to
650 the \refa{Extract Class} refactoring, and only allows for \emph{fields} to be
651 moved to a new class, \emph{not methods}. This makes it effectively only
652 extracting a data structure, and calling it \refa{Extract Class} is a little
653 misleading. One would often be better off with textual extract and paste than
654 using the \refa{Extract Class} refactoring in \name{Eclipse}. When it comes to
655 \name{NetBeans}, it does not even show an attempt on providing this refactoring.
657 \subsection{The relation to design patterns}\label{relationToDesignPatterns}
659 Refactoring and design patterns have at least one thing in common, they are both
660 promoted by advocates of \emph{clean code}\citing{cleanCode} as fundamental
661 tools on the road to more maintainable and extendable source code.
664 Design patterns help you determine how to reorganize a design, and they can
665 reduce the amount of refactoring you need to do
666 later.~\cite[p.~353]{designPatterns}
669 Although sometimes associated with
670 over-engineering\citing{kerievsky2005,refactoring}, design patterns are in
671 general assumed to be good for maintainability of source code. That may be
672 because many of them are designed to support the \emph{open/closed principle} of
673 object-oriented programming. The principle was first formulated by Bertrand
674 Meyer, the creator of the Eiffel programming language, like this: ``Modules
675 should be both open and closed.''\citing{meyer1988} It has been popularized,
676 with this as a common version:
679 Software entities (classes, modules, functions, etc.) should be open for
680 extension, but closed for modification.
683 Maintainability is often thought of as the ability to be able to introduce new
684 functionality without having to change too much of the old code. When
685 refactoring, the motivation is often to facilitate adding new functionality. It
686 is about factoring the old code in a way that makes the new functionality being
687 able to benefit from the functionality already residing in a software system,
688 without having to copy old code into new. Then, next time someone shall add new
689 functionality, it is less likely that the old code has to change. Assuming that
690 a design pattern is the best way to get rid of duplication and assist in
691 implementing new functionality, it is reasonable to conclude that a design
692 pattern often is the target of a series of refactorings. Having a repertoire of
693 design patterns can also help in knowing when and how to refactor a program to
694 make it reflect certain desired characteristics.
697 There is a natural relation between patterns and refactorings. Patterns are
698 where you want to be; refactorings are ways to get there from somewhere
699 else.~\cite[p.~107]{refactoring}
702 This quote is wise in many contexts, but it is not always appropriate to say
703 ``Patterns are where you want to be\ldots''. \emph{Sometimes}, patterns are
704 where you want to be, but only because it will benefit your design. It is not
705 true that one should always try to incorporate as many design patterns as
706 possible into a program. It is not like they have intrinsic value. They only add
707 value to a system when they support its design. Otherwise, the use of design
708 patterns may only lead to a program that is more complex than necessary.
711 The overuse of patterns tends to result from being patterns happy. We are
712 \emph{patterns happy} when we become so enamored of patterns that we simply
713 must use them in our code.~\cite[p.~24]{kerievsky2005}
716 This can easily happen when relying largely on up-front design. Then it is
717 natural, in the very beginning, to try to build in all the flexibility that one
718 believes will be necessary throughout the lifetime of a software system.
719 According to Joshua Kerievsky ``That sounds reasonable --- if you happen to be
720 psychic.''~\cite[p.~1]{kerievsky2005} He is advocating what he believes is a
721 better approach: To let software continually evolve. To start with a simple
722 design that meets today's needs, and tackle future needs by refactoring to
723 satisfy them. He believes that this is a more economic approach than investing
724 time and money into a design that inevitably is going to change. By relying on
725 continuously refactoring a system, its design can be made simpler without
726 sacrificing flexibility. To be able to fully rely on this approach, it is of
727 utter importance to have a reliable suit of tests to lean on \see{testing}. This
728 makes the design process more natural and less characterized by difficult
729 decisions that has to be made before proceeding in the process, and that is
730 going to define a project for all of its unforeseeable future.
732 \subsection{The impact on software quality}
734 \subsubsection{What is software quality?}
735 The term \emph{software quality} has many meanings. It all depends on the
736 context we put it in. If we look at it with the eyes of a software developer, it
737 usually means that the software is easily maintainable and testable, or in other
738 words, that it is \emph{well designed}. This often correlates with the
739 management scale, where \emph{keeping the schedule} and \emph{customer
740 satisfaction} is at the center. From the customers point of view, in addition to
741 good usability, \emph{performance} and \emph{lack of bugs} is always
742 appreciated, measurements that are also shared by the software developer. (In
743 addition, such things as good documentation could be measured, but this is out
744 of the scope of this document.)
746 \subsubsection{The impact on performance}
748 Refactoring certainly will make software go more slowly\footnote{With today's
749 compiler optimization techniques and performance tuning of e.g. the Java
750 virtual machine, the penalties of object creation and method calls are
751 debatable.}, but it also makes the software more amenable to performance
752 tuning.~\cite[p.~69]{refactoring}
755 \noindent There is a common belief that refactoring compromises performance, due
756 to increased degree of indirection and that polymorphism is slower than
759 In a survey, Demeyer\citing{demeyer2002} disproves this view in the case of
760 polymorphism. He did an experiment on, what he calls, ``Transform Self Type
761 Checks'' where you introduce a new polymorphic method and a new class hierarchy
762 to get rid of a class' type checking of a ``type attribute``. He uses this kind
763 of transformation to represent other ways of replacing conditionals with
764 polymorphism as well. The experiment is performed on the C++ programming
765 language and with three different compilers and platforms. Demeyer concludes
766 that, with compiler optimization turned on, polymorphism beats middle to large
767 sized if-statements and does as well as case-statements. (In accordance with
768 his hypothesis, due to similarities between the way C++ handles polymorphism and
772 The interesting thing about performance is that if you analyze most programs,
773 you find that they waste most of their time in a small fraction of the
774 code.~\cite[p.~70]{refactoring}
777 \noindent So, although an increased amount of method calls could potentially
778 slow down programs, one should avoid premature optimization and sacrificing good
779 design, leaving the performance tuning until after \gloss{profiling} the
780 software and having isolated the actual problem areas.
782 \subsection{Composite refactorings}\label{compositeRefactorings}
783 Generally, when thinking about refactoring, at the mechanical level, there are
784 essentially two kinds of refactorings. There are the \emph{primitive}
785 refactorings, and the \emph{composite} refactorings.
787 \definition{A \emph{primitive refactoring} is a refactoring that cannot be
788 expressed in terms of other refactorings.}
790 \noindent Examples are the \refa{Pull Up Field} and \refa{Pull Up
791 Method} refactorings\citing{refactoring}, that move members up in their class
794 \definition{A \emph{composite refactoring} is a refactoring that can be
795 expressed in terms of two or more other refactorings.}
797 \noindent An example of a composite refactoring is the \refa{Extract
798 Superclass} refactoring\citing{refactoring}. In its simplest form, it is composed
799 of the previously described primitive refactorings, in addition to the
800 \refa{Pull Up Constructor Body} refactoring\citing{refactoring}. It works
801 by creating an abstract superclass that the target class(es) inherits from, then
802 by applying \refa{Pull Up Field}, \refa{Pull Up Method} and
803 \refa{Pull Up Constructor Body} on the members that are to be members of
804 the new superclass. If there are multiple classes in play, their interfaces may
805 need to be united with the help of some rename refactorings, before extracting
806 the superclass. For an overview of the \refa{Extract Superclass}
807 refactoring, see \myref{fig:extractSuperclass}.
811 \includegraphics[angle=270,width=\linewidth]{extractSuperclassItalic.pdf}
812 \caption{The Extract Superclass refactoring, with united interfaces.}
813 \label{fig:extractSuperclass}
816 \subsection{Manual vs. automated refactorings}
817 Refactoring is something every programmer does, even if \heshe does not known
818 the term \emph{refactoring}. Every refinement of source code that does not alter
819 the program's behavior is a refactoring. For small refactorings, such as
820 \ExtractMethod, executing it manually is a manageable task, but is still prone
821 to errors. Getting it right the first time is not easy, considering the method
822 signature and all the other aspects of the refactoring that has to be in place.
824 Consider the renaming of classes, methods and fields. For complex programs these
825 refactorings are almost impossible to get right. Attacking them with textual
826 search and replace, or even regular expressions, will fall short on these tasks.
827 Then it is crucial to have proper tool support that can perform them
828 automatically. Tools that can parse source code and thus have semantic knowledge
829 about which occurrences of which names belong to what construct in the program.
830 For even trying to perform one of these complex tasks manually, one would have
831 to be very confident on the existing test suite \see{testing}.
833 \subsection{Correctness of refactorings}\label{sec:correctness}
834 For automated refactorings to be truly useful, they must show a high degree of
835 behavior preservation. This last sentence might seem obvious, but there are
836 examples of refactorings in existing tools that break programs. In an ideal
837 world, every automated refactoring would be ``complete'', in the sense that it
838 would never break a program. In an ideal world, every program would also be free
839 from bugs. In modern IDEs the implemented automated refactorings are working for
840 \emph{most} cases, which is enough for making them useful.
842 I will now present an example of a \emph{corner case} where a program breaks
843 when a refactoring is applied. The example shows an \ExtractMethod refactoring
844 followed by a \MoveMethod refactoring that breaks a program in both the
845 \name{Eclipse} and \name{IntelliJ} IDEs\footnote{The \name{NetBeans} IDE handles this
846 particular situation without altering the program's behavior, mainly because
847 its \refa{Move Method} refactoring implementation is a bit flawed in other ways
848 \see{toolSupport}.}. The target and the destination for the composed
849 refactoring are shown in \myref{lst:correctnessExtractAndMove}. Note that the
850 method \method{m(C c)} of class \type{X} assigns to the field \var{x} of the
851 argument \var{c} that has type \type{C}.
855 \begin{minted}[linenos,frame=topline,label={Refactoring
856 target},framesep=\mintedframesep]{java}
858 public X x = new X();
870 \begin{minted}[frame=topline,label={Method
871 destination},framesep=\mintedframesep]{java}
875 // If m is called from
876 // c, then c.x no longer
883 \caption{The target and the destination for the composition of the Extract
884 Method and \refa{Move Method} refactorings.}
885 \label{lst:correctnessExtractAndMove}
889 The refactoring sequence works by extracting line 6 through 8 from the original
890 class \type{C} into a method \method{f} with the statements from those lines as
891 its method body (but with the comment left out, since it will no longer hold any
892 meaning). The method is then moved to the class \type{X}. The result is shown
893 in \myref{lst:correctnessExtractAndMoveResult}.
895 Before the refactoring, the methods \method{m} and \method{n} of class \type{X}
896 are called on different object instances (see line 6 and 8 of the original class
897 \type{C} in \cref{lst:correctnessExtractAndMove}). After the refactoring, they
898 are called on the same object, and the statement on line
899 3 of class \type{X} (in \cref{lst:correctnessExtractAndMoveResult}) no longer
900 has the desired effect in our example. The method \method{f} of class \type{C}
901 is now calling the method \method{f} of class \type{X} (see line 5 of class
902 \type{C} in \cref{lst:correctnessExtractAndMoveResult}), and the program now
903 behaves different than before.
907 \begin{minted}[linenos]{java}
909 public X x = new X();
919 \begin{minted}[linenos]{java}
934 \caption{The result of the composed refactoring.}
935 \label{lst:correctnessExtractAndMoveResult}
938 The bug introduced in the previous example is of such a nature\footnote{Caused
939 by aliasing.} that it is very difficult to spot if the refactored code is not
940 covered by tests. It does not generate compilation errors, and will thus only
941 result in a runtime error or corrupted data, which might be hard to detect.
943 \subsection{Refactoring and the importance of testing}\label{testing}
945 If you want to refactor, the essential precondition is having solid
946 tests.\citing{refactoring}
949 When refactoring, there are roughly three classes of errors that can be made.
950 The first class of errors is the one that makes the code unable to compile.
951 These \emph{compile-time} errors are of the nicer kind. They flash up at the
952 moment they are made (at least when using an IDE), and are usually easy to fix.
953 The second class is the \emph{runtime} errors. Although these errors take a bit
954 longer to surface, they usually manifest after some time in an illegal argument
955 exception, null pointer exception or similar during the program execution.
956 These kinds of errors are a bit harder to handle, but at least they will show,
957 eventually. Then there are the \emph{behavior-changing} errors. These errors are
958 of the worst kind. They do not show up during compilation and they do not turn
959 on a blinking red light during runtime either. The program can seem to work
960 perfectly fine with them in play, but the business logic can be damaged in ways
961 that will only show up over time.
963 For discovering runtime errors and behavior changes when refactoring, it is
964 essential to have good test coverage. Testing in this context means writing
965 automated tests. Manual testing may have its uses, but when refactoring, it is
966 automated unit testing that dominate. For discovering behavior changes it is
967 especially important to have tests that cover potential problems, since these
968 kinds of errors do not reveal themselves.
970 Unit testing is not a way to \emph{prove} that a program is correct, but it is a
971 way to make you confident that it \emph{probably} works as desired. In the
972 context of test-driven development (commonly known as TDD), the tests are even a
973 way to define how the program is \emph{supposed} to work. It is then, by
974 definition, working if the tests are passing.
976 If the test coverage for a code base is perfect, then it should, theoretically,
977 be risk-free to perform refactorings on it. This is why automated tests and
978 refactoring is such a great match.
980 \subsubsection{Testing the code from correctness section}
981 The worst thing that can happen when refactoring is to introduce changes to the
982 behavior of a program, as in the example on \myref{sec:correctness}. This
983 example may be trivial, but the essence is clear. The only problem with the
984 example is that it is not clear how to create automated tests for it, without
985 changing it in intrusive ways.
987 Unit tests, as they are known from the different \glosspl{xUnit} around, are
988 only suitable to test the \emph{result} of isolated operations. They can not
989 easily (if at all) observe the \emph{history} of a program.
991 This problem is still open.
995 Assuming a sequential (non-concurrent) program:
998 tracematch (C c, X x) {
1000 call(* X.m(C)) && args(c) && cflow(within(C));
1002 call(* X.n()) && target(x) && cflow(within(C));
1004 set(C.x) && target(c) && !cflow(m);
1008 { assert x == c.x; }
1012 %\begin{minted}{java}
1013 %tracematch (X x1, X x2) {
1015 % call(* X.m(C)) && target(x1);
1017 % call(* X.n()) && target(x2);
1019 % set(C.x) && !cflow(m) && !cflow(n);
1023 % { assert x1 != x2; }
1029 \section{The project}\label{sec:project}
1030 In this section we look at the work that shall be done for this project, its
1031 building stones and some of the methodologies used.
1033 \subsection{Project description}
1034 The aim of this master's project will be to explore the relationship between the
1035 \ExtractMethod and the \MoveMethod refactorings. This will be done by composing
1036 the two into a composite refactoring. The refactoring will be called the
1037 \ExtractAndMoveMethod refactoring.
1039 The two primitive \ExtractMethod and \MoveMethod refactorings must already be
1040 implemented in a tool, so the \ExtractAndMoveMethod refactoring is going to be
1041 built on top of those.
1043 The composition of the \ExtractMethod and \MoveMethod refactorings springs
1044 naturally out of the need to move procedures closer to the data they manipulate.
1045 This composed refactoring is not well described in the literature, but it is
1046 implemented in at least one tool called
1047 \name{CodeRush}\footnote{\url{https://help.devexpress.com/\#CodeRush/CustomDocument3519}},
1048 which is an extension for \name{MS Visual
1049 Studio}\footnote{\url{http://www.visualstudio.com/}}. In CodeRush it is called
1050 \refa{Extract Method to
1051 Type}\footnote{\url{https://help.devexpress.com/\#CodeRush/CustomDocument6710}},
1052 but I choose to call it \ExtractAndMoveMethod, since I feel this better
1053 communicates which primitive refactorings it is composed of.
1055 The project will consist of implementing the \ExtractAndMoveMethod refactoring,
1056 as well as executing it over a larger code base, as a case study. To be able to
1057 execute the refactoring automatically, I have to make it analyze code to
1058 determine the best selections to extract into new methods.
1060 \subsection{The premises}
1061 Before we can start manipulating source code and write a tool for doing so, we
1062 need to decide on a programming language for the code we are going to
1063 manipulate. Also, since we do not want to start from scratch by implementing
1064 primitive refactorings ourselves, we need to choose an existing tool that
1065 provides the needed refactorings. In addition to be able to perform changes, we
1066 need a framework for analyzing source code for the language we select.
1068 \subsubsection{Choosing the target language}
1069 Choosing which programming language the code that shall be manipulated shall be
1070 written in, is not a very difficult task. We choose to limit the possible
1071 languages to the object-oriented programming languages, since most of the
1072 terminology and literature regarding refactoring comes from the world of
1073 object-oriented programming. In addition, the language must have existing tool
1074 support for refactoring.
1076 The \name{Java} programming language\footnote{\url{https://www.java.com/}} is
1077 the dominating language when it comes to example code in the literature of
1078 refactoring, and is thus a natural choice. Java is perhaps, currently the most
1079 influential programming language in the world, with its \name{Java Virtual
1080 Machine} that runs on all of the most popular architectures and also supports
1081 dozens of other programming languages\footnote{They compile to Java bytecode.},
1082 with \name{Scala}, \name{Clojure} and \name{Groovy} as the most prominent ones.
1083 Java is currently the language that every other programming language is compared
1084 against. It is also the primary programming language for the author of this
1087 \subsubsection{Choosing the tools}
1088 When choosing a tool for manipulating Java, there are certain criteria that
1089 have to be met. First of all, the tool should have some existing refactoring
1090 support that this thesis can build upon. Secondly it should provide some kind of
1091 framework for parsing and analyzing Java source code. Third, it should itself be
1092 open source. This is both because of the need to be able to browse the code for
1093 the existing refactorings that is contained in the tool, and also because open
1094 source projects hold value in them selves. Another important aspect to consider
1095 is that open source projects of a certain size, usually has large communities of
1096 people connected to them, that are committed to answering questions regarding the
1097 use and misuse of the products, that to a large degree is made by the community
1100 There is a certain class of tools that meet these criteria, namely the class of
1101 \emph{IDEs}\footnote{\emph{Integrated Development Environment}}. These are
1102 programs that are meant to support the whole production cycle of a computer
1103 program, and the most popular IDEs that support Java, generally have quite good
1104 refactoring support.
1106 The main contenders for this thesis is the \name{Eclipse IDE}, with the
1107 \name{Java development tools} (JDT), the \name{IntelliJ IDEA Community Edition}
1108 and the \name{NetBeans IDE} \see{toolSupport}. \name{Eclipse} and
1109 \name{NetBeans} are both free, open source and community driven, while the
1110 \name{IntelliJ IDEA} has an open sourced community edition that is free of
1111 charge, but also offer an \name{Ultimate Edition} with an extended set of
1112 features, at additional cost. All three IDEs supports adding plugins to extend
1113 their functionality and tools that can be used to parse and analyze Java source
1114 code. But one of the IDEs stand out as a favorite, and that is the \name{Eclipse
1115 IDE}. This is the most popular\citing{javaReport2011} among them and seems to be
1116 de facto standard IDE for Java development regardless of platform.
1119 \subsection{The primitive refactorings}
1120 The refactorings presented here are the primitive refactorings used in this
1121 project. They are the abstract building blocks used by the \ExtractAndMoveMethod
1124 \paragraph{The Extract Method refactoring}
1125 The \refa{Extract Method} refactoring is used to extract a fragment of code
1126 from its context and into a new method. A call to the new method is inlined
1127 where the fragment was before. It is used to break code into logical units, with
1128 names that explain their purpose.
1130 An example of an \ExtractMethod refactoring is shown in
1131 \myref{lst:extractMethodRefactoring}. It shows a method containing calls to the
1132 methods \method{foo} and \method{bar} of a type \type{X}. These statements are
1133 then extracted into the new method \method{fooBar}.
1136 \begin{multicols}{2}
1137 \begin{minted}[samepage,frame=topline,label={Before},framesep=\mintedframesep]{java}
1148 \begin{minted}[samepage,frame=topline,label={After},framesep=\mintedframesep]{java}
1160 \caption{An example of an \ExtractMethod refactoring.}
1161 \label{lst:extractMethodRefactoring}
1164 \paragraph{The Move Method refactoring}
1165 The \refa{Move Method} refactoring is used to move a method from one class to
1166 another. This can be appropriate if the method is using more features of another
1167 class than of the class which it is currently defined.
1169 \Myref{lst:moveMethodRefactoring} shows an example of this refactoring. Here a
1170 method \method{fooBar} is moved from the class \type{C} to the class \type{X}.
1173 \begin{multicols}{2}
1174 \begin{minted}[samepage,frame=topline,label={Before},framesep=\mintedframesep]{java}
1193 \begin{minted}[samepage,frame=topline,label={After},framesep=\mintedframesep]{java}
1210 \caption{An example of a \MoveMethod refactoring.}
1211 \label{lst:moveMethodRefactoring}
1214 \subsection{The Extract and Move Method refactoring}
1215 The \ExtractAndMoveMethod refactoring is a composite refactoring composed of the
1216 primitive \ExtractMethod and \MoveMethod refactorings. The effect of this
1217 refactoring on source code is the same as when extracting a method and moving it
1218 to another class. Conceptually, this is done without an intermediate step. In
1219 practice, as we shall see later, an intermediate step may be necessary.
1221 An example of this composite refactoring is shown in
1222 \myref{lst:extractAndMoveMethodRefactoring}. The example joins the examples from
1223 \cref{lst:extractMethodRefactoring} and \cref{lst:moveMethodRefactoring}. This
1224 means that the selection consisting of the consecutive calls to the methods
1225 \method{foo} and \method{bar}, is extracted into a new method \method{fooBar}
1226 located in the class \type{X}.
1229 \begin{multicols}{2}
1230 \begin{minted}[samepage,frame=topline,label={Before},framesep=\mintedframesep]{java}
1246 \begin{minted}[samepage,frame=topline,label={After},framesep=\mintedframesep]{java}
1263 \caption{An example of the \ExtractAndMoveMethod refactoring.}
1264 \label{lst:extractAndMoveMethodRefactoring}
1267 \subsection{The Coupling Between Object Classes metric}\label{sec:CBO}
1268 The best known metric for measuring coupling between classes in object-oriented
1269 software is called \metr{Coupling Between Object Classes}, usually abbreviated
1270 as CBO. The metric is defined in the article \tit{A Metrics Suite for Object
1271 Oriented Design}\citing{metricsSuite1994} by Chidamber and Kemerer, published in
1274 \definition{\emph{CBO} for a class is a count of the number of other classes to
1275 which it is coupled.}
1277 An object is coupled to another object if one of them acts on the other by using
1278 methods or instance variables of the other object. This relation goes both ways,
1279 so both outgoing and incoming uses are counted. Each coupling relationship is
1280 only considered once when measuring CBO for a class.
1282 \paragraph{How can the Extract and Move Method refactoring improve CBO?}
1283 \Myref{lst:CBOExample} shows how CBO changes for a class when it is refactored
1284 with the \ExtractAndMoveMethod refactoring. In the example we consider only the
1285 CBO value of class \type{C}.
1288 \begin{multicols}{2}
1289 \begin{minted}[linenos,samepage,frame=topline,label={Before},framesep=\mintedframesep]{java}
1320 \begin{minted}[linenos,samepage,frame=topline,label={After},framesep=\mintedframesep]{java}
1352 \caption{An example of improving CBO. Class \type{C} has a CBO value of 4
1353 before refactoring it, and 3 after.}
1354 \label{lst:CBOExample}
1357 Before refactoring the class \type{C} with the \ExtractAndMoveMethod
1358 refactoring, it has a CBO value of 4. The class uses members of the classes
1359 \type{A} and \type{B}, which accounts for 2 of the coupling relationships of
1360 class \type{C}. In addition to this, it uses its variable \var{x} with type
1361 \type{X} and also the methods \method{foo} and \method{bar} declared in class
1362 \type{Y}, giving it a total CBO value of 4.
1364 The after-part of the example code in \mysimpleref{lst:CBOExample} shows the
1365 result of extracting the lines
1366 5 and 6 of class \type{C} into a new method \method{fooBar}, with a subsequent
1367 move of it to class \type{X}.
1369 With respect to the CBO metric, the refactoring action accomplishes something
1370 important: It eliminates the uses of class \type{Y} from class \type{C}. This
1371 means that the class \type{C} is no longer coupled to \type{Y}, only the classes
1372 \type{A}, \type{B} and \type{X}. The CBO value of class \type{C} is therefore 3
1373 after refactoring, while no other class have received any increase in CBO.
1375 The example shown here is an ideal situation. Coupling is reduced for one class
1376 without any increase of coupling for another class. There is also another point
1377 that is important. It is the fact that to reduce the CBO value for a class, we
1378 need to remove \emph{all} its uses of another class. This is done for the class
1379 \type{C} in \myref{lst:CBOExample}, where all uses of class \type{Y} is removed
1380 by the \ExtractAndMoveMethod refactoring.
1381 \todoin{Highlight code}
1384 \subsection{Research questions}\label{sec:researchQuestions}
1385 The main question that I seek an answer to in this thesis is:
1388 Is it possible to automate the analysis and execution of the
1389 \ExtractAndMoveMethod refactoring, and do so for all of the code of a larger
1393 \noindent The secondary questions will then be:
1395 \paragraph{Can we do this efficiently?} Can we automate the analysis and
1396 execution of the refactoring so it can be run in a reasonable amount of time?
1398 \paragraph{Can we perform changes safely?} Can we take actions to prevent the
1399 refactoring from breaking the code? By breaking the code we mean to either do
1400 changes that do not compile, or make changes that alter the behavior of the
1403 \paragraph{Can we improve the quality of source code?} Assuming that the
1404 refactoring is safe: Is it feasible to assure that the code we refactor actually
1405 gets better in terms of coupling?
1407 \paragraph{How can the automation of the refactoring be helpful?} Assuming the
1408 refactoring does in fact improve the quality of source code and is safe to use:
1409 What is the usefulness of the refactoring in a software development setting? In
1410 what parts of the development process can the refactoring play a role?
1412 \subsection{Methodology}
1413 This section will present some of the methods used during the work of this
1416 \subsubsection{Evolutionary design}
1417 In the programming work for this project, I have tried using a design strategy
1418 called evolutionary design, also known as continuous or incremental
1419 design\citing{wiki_continuous_2014}. It is a software design strategy advocated
1420 by the Extreme Programming community. The essence of the strategy is that you
1421 should let the design of your program evolve naturally as your requirements
1422 change. This is seen in contrast with up-front design, where design decisions
1423 are made early in the process.
1425 The motivation behind evolutionary design is to keep the design of software as
1426 simple as possible. This means not introducing unneeded functionality into a
1427 program. You should defer introducing flexibility into your software, until it
1428 is needed to be able to add functionality in a clean way.
1430 Holding up design decisions, implies that the time will eventually come when
1431 decisions have to be made. The flexibility of the design then relies on the
1432 programmer's abilities to perform the necessary refactoring, and \his confidence
1433 in those abilities. From my experience working on this project, I can say that
1434 this confidence is greatly enhanced by having automated tests to rely on
1437 The choice of going for evolutionary design developed naturally. As Fowler
1438 points out in his article \tit{Is Design Dead?}, evolutionary design much
1439 resembles the ``code and fix'' development strategy\citing{fowler_design_2004}.
1440 A strategy that most of us have practiced in school. This was also the case when
1441 I first started this work. I had to learn the inner workings of Eclipse and its
1442 refactoring-related plugins. That meant a lot of fumbling around with code I did
1443 not know, in a trial and error fashion. Eventually I started writing tests for
1444 my code, and my design began to evolve.
1446 \subsubsection{Test-driven development}\label{tdd}
1447 As mentioned before, the project started out as a classic code and fix
1448 development process. My focus was aimed at getting something to work, rather
1449 than doing so according to best practice. This resulted in a project that got
1450 out of its starting blocks, but it was not accompanied by any tests. Hence it
1451 was soon difficult to make any code changes with the confidence that the program
1452 was still correct afterwards (assuming it was so before changing it). I always
1453 knew that I had to introduce some tests at one point, but this experience
1454 accelerated the process of leading me onto the path of testing.
1456 I then wrote tests for the core functionality of the plugin, and thus gained
1457 more confidence in the correctness of my code. I could now perform quite drastic
1458 changes without ``wetting my pants``. After this, nearly all of the semantic
1459 changes done to the business logic of the project, or the addition of new
1460 functionality, were made in a test-driven manner. This means that before
1461 performing any changes, I would define the desired functionality through a set
1462 of tests. I would then run the tests to check that they were run and that they
1463 did not pass. Then I would do any code changes necessary to make the tests
1464 pass. The definition of how the program is supposed to operate is then captured
1465 by the tests. However, this does not prove the correctness of the analysis
1466 leading to the test definitions.
1468 \subsection{Case studies}
1469 The case study methodology is used to show how the \ExtractAndMoveMethod
1470 refactoring performs on real code, not just toy examples. The case studies are
1471 used to analyze our project so we can conclude on its completeness and
1474 \subsection{Dogfooding}
1475 Dogfooding is a methodology where you use your own tools to do your job, also
1476 referred to as ``eating your own dog food''\citing{harrisonDogfooding2006}. It
1477 is used in this project to see if we can refactor our own refactoring code and
1478 still use it to refactor other code.
1480 \section{Related work}\label{sec:relatedWork}
1481 Here we present some work related to automated composition of refactorings.
1483 \subsection{``Making Program Refactoring
1484 Safer''}\label{sec:saferRefactoringTests}
1485 This is the name of an article\citing{soaresSafer2010} about providing a way to
1486 improve safety during refactoring. Soares et al. approaches the problem of
1487 preserving behavior during refactoring by analyzing a transformation and then
1488 generate a test suite for it, using static analysis. These tests are then run
1489 for both the before- and after-code, and is compared to assure that they are
1492 \subsection{Search-based refactoring}
1493 \tit{Search-Based Refactoring: an
1494 empirical study}\citing{okeeffeSearchBased2008} is a paper by Mark O'Keeffe and
1495 Mel Ó Cinnéide published in 2008. The authors present an empirical study of
1496 different algorithmic approaches to search-based refactoring.
1498 The common approach for all these algorithms is to generate a set of changes to
1499 a program for then to use a ``fitness function'' to evaluate if they improve its
1500 design or not. The fitness function consists of a weighted sum of different
1501 object-oriented metrics.
1503 Among other things, the authors conclude that even with small input programs,
1504 their solution representation is memory-intensive, at least for some algorithms.
1505 The programs they refactor on have in average 4,000 lines of code, spread over
1506 57 classes. I.e. considerably smaller than one of the programs that will be
1507 subject to refactoring in this project.
1510 \subsection{The compositional paradigm of refactoring}
1511 This paradigm builds upon the observation of Vakilian et
1512 al.\citing{vakilian2012}, that of the many automated refactorings existing in
1513 modern IDEs, the simplest ones are dominating the usage statistics. The report
1514 mainly focuses on \name{Eclipse} as the tool under investigation.
1516 The paradigm is described almost as the opposite of automated composition of
1517 refactorings \see{compositeRefactorings}. It works by providing the programmer
1518 with easily accessible primitive refactorings. These refactorings shall be
1519 accessed via keyboard shortcuts or quick-assist menus\footnote{Think
1520 quick-assist with Ctrl+1 in \name{Eclipse}} and be promptly executed, opposed to in the
1521 currently dominating wizard-based refactoring paradigm. They are meant to
1522 stimulate composing smaller refactorings into more complex changes, rather than
1523 doing a large upfront configuration of a wizard-based refactoring, before
1524 previewing and executing it. The compositional paradigm of refactoring is
1525 supposed to give control back to the programmer, by supporting \himher with an
1526 option of performing small rapid changes instead of large changes with a lesser
1527 degree of control. The report authors hope this will lead to fewer unsuccessful
1528 refactorings. It also could lower the bar for understanding the steps of a
1529 larger composite refactoring and thus also help in figuring out what goes wrong
1530 if one should choose to op in on a wizard-based refactoring.
1532 Vakilian and his associates have performed a survey of the effectiveness of the
1533 compositional paradigm versus the wizard-based one. They claim to have found
1534 evidence of that the \emph{compositional paradigm} outperforms the
1535 \emph{wizard-based}. It does so by reducing automation, which seems
1536 counterintuitive. Therefore they ask the question ``What is an appropriate level
1537 of automation?'', and thus questions what they feel is a rush toward more
1538 automation in the software engineering community.
1542 \chapter{The search-based Extract and Move Method
1543 refactoring}\label{ch:extractAndMoveMethod}
1544 In this chapter I will delve into the workings of the search-based
1545 \ExtractAndMoveMethod refactoring. We will see the choices it must make along
1546 the way and why it chooses a text selection as a candidate for refactoring or
1549 After defining some concepts, I will introduce an example that will be used
1550 throughout the chapter to illustrate how the refactoring works in some simple
1553 \section{The inputs to the refactoring}
1554 For executing an \ExtractAndMoveMethod refactoring, there are two simple
1555 requirements. The first thing the refactoring needs is a text selection, telling
1556 it what to extract. Its second requirement is a target for the subsequent move
1559 The extracted method must be called instead of the selection that makes up its
1560 body. Also, the method call has to be performed via a variable, since the method
1561 is not static. Therefore, the move target must be a variable in the scope of the
1562 extracted selection. The actual new location for the extracted method will be
1563 the class representing the type of the move target variable. But, since the
1564 method also must be called through a variable, it makes sense to define the move
1565 target to be either a local variable or a field in the scope of the text
1568 \section{Defining a text selection}
1569 A text selection, in our context, is very similar to what you think of when
1570 selecting a bit of text in your editor or other text processing tool with your
1571 mouse or keyboard. It is an abstract construct that is meant to capture which
1572 specific portion of text we are about to deal with.
1574 To be able to clearly reason about a text selection done to a portion of text in
1575 a computer file, which consists of pure text, we put up the following
1578 \definition{A \emph{text selection} in a text file is defined by two
1579 non-negative integers, in addition to a reference to the file itself. The first
1580 integer is an offset into the file, while the second reference is the length of
1581 the text selection.}
1583 This means that the selected text consist of a number of characters equal to the
1584 length of the selection, where the first character is found at the specified
1587 \section{Where we look for text selections}
1589 \subsection{Text selections are found in methods}
1590 The text selections we are interested in are those that surround program
1591 statements. Therefore, the place we look for selections that can form candidates
1592 for an execution of the \ExtractAndMoveMethod refactoring, is within the body of
1595 \paragraph{On ignoring static methods}
1596 In this project we are not analyzing static methods for candidates to the
1597 \ExtractAndMoveMethod refactoring. The reason for this is that in the cases
1598 where we want to perform the refactoring for a selection within a static method,
1599 the first step is to extract the selection into a new method. Hence this method
1600 also becomes static, since it must be possible to call it from a static context.
1601 It would then be difficult to move the method to another class, make it
1602 non-static and calling it through a variable. To avoid these obstacles, we
1603 simply ignore static methods.
1605 \begin{listing}[htb]
1606 \def\charwidth{5.8pt}
1607 \def\indent{2*\charwidth}
1608 \def\lineheight{\baselineskip}
1609 \def\mintedtop{2*\lineheight+5.8pt}
1611 \begin{tikzpicture}[overlay, yscale=-1, xshift=3.8pt+\charwidth*31]
1612 \tikzstyle{overlaybox}=[fill=lightgray,opacity=0.2]
1614 \draw[overlaybox] (\indent,\mintedtop+\lineheight*4) rectangle
1615 +(23*\charwidth,17*\lineheight);
1618 \draw[overlaybox] (2*\indent,\mintedtop+5*\lineheight) rectangle
1619 +(15*\charwidth,3*\lineheight);
1620 \draw[overlaybox] (2*\indent,\mintedtop+15*\lineheight) rectangle
1621 +(15*\charwidth,3*\lineheight);
1622 \draw[overlaybox] (2*\indent,\mintedtop+19*\lineheight) rectangle
1623 +(15*\charwidth,\lineheight);
1625 \begin{multicols}{2}
1626 \begin{minted}[linenos,frame=topline,label=Clean,framesep=\mintedframesep]{java}
1628 A a; B b; boolean bool;
1630 void method(int val) {
1654 \begin{minted}[frame=topline,label={With statement
1655 sequences},framesep=\mintedframesep]{java}
1657 A a; B b; boolean bool;
1659 void method(int val) {
1682 \caption{Classes \type{A} and \type{B} are both public. The methods
1683 \method{foo} and \method{bar} are public members of class \type{A}.}
1684 \label{lst:grandExample}
1687 \subsection{The possible text selections of a method body}
1688 The number of possible text selections that can be made from the text in a
1689 method body, are equal to all the sub-sequences of characters within it. For our
1690 purposes, analyzing program source code, we must define what it means for a text
1691 selection to be valid.
1693 \definition{A \emph{valid text selection} is a text selection that contains all
1694 of one or more consecutive program statements.}
1696 For a sequence of statements, the text selections that can be made from it, are
1697 equal to all its sub-sequences. \Myref{lst:textSelectionsExample} show an
1698 example of all the text selections that can be made from the code in
1699 \myref{lst:grandExample}, lines 16-18. For convenience and the clarity of this
1700 example, the text selections are represented as tuples with the start and end
1701 line of all selections: $\{(16), (17), (18), (16,17), (16,18), (17,18)\}$.
1703 \begin{listing}[htb]
1704 \def\charwidth{5.7pt}
1705 \def\indent{4*\charwidth}
1706 \def\lineheight{\baselineskip}
1707 \def\mintedtop{\lineheight-1pt}
1709 \begin{tikzpicture}[overlay, yscale=-1]
1710 \tikzstyle{overlaybox}=[fill=lightgray,opacity=0.2]
1713 \draw[overlaybox] (2*\charwidth,\mintedtop) rectangle
1714 +(16*\charwidth,\lineheight);
1717 \draw[overlaybox] (2*\charwidth,\mintedtop+\lineheight) rectangle
1718 +(16*\charwidth,\lineheight);
1721 \draw[overlaybox] (2*\charwidth,\mintedtop+2*\lineheight) rectangle
1722 +(16*\charwidth,\lineheight);
1724 \draw[overlaybox] (\indent-3*\charwidth,\mintedtop) rectangle
1725 +(18*\charwidth,2*\lineheight);
1727 \draw[overlaybox] (3*\charwidth,\mintedtop+\lineheight) rectangle
1728 +(14*\charwidth,2*\lineheight);
1731 \draw[overlaybox] (\indent,\mintedtop) rectangle
1732 +(12*\charwidth,3*\lineheight);
1734 % indent should be 5 spaces
1735 \begin{minted}[linenos,firstnumber=16]{java}
1740 \caption{Example of how the text selections generator would generate text
1741 selections based on a lists of statements. Each highlighted rectangle
1742 represents a text selection.}
1743 \label{lst:textSelectionsExample}
1746 Each nesting level of a method body can have many such sequences of statements.
1747 The outermost nesting level has one such sequence, and each branch contains
1748 its own sequence of statements. \Myref{lst:grandExample} has a version of some
1749 code where all such sequences of statements are highlighted for a method body.
1751 To complete our example of possible text selections, I will now list all
1752 possible text selections for the method in \myref{lst:grandExample}, by nesting
1753 level. There are 23 of them in total.
1756 \item[Level 1 (10 selections)] \hfill \\
1757 $\{(5,9), (11), (12), (14,21), (5,11), (5,12), (5,21), (11,12),
1758 (11,21), \\(12,21)\}$
1760 \item[Level 2 (13 selections)] \hfill \\
1761 $\{(6), (7), (8), (6,7), (6,8), (7,8), (16), (17), (18), (16,17), (16,18), \\
1765 \subsubsection{The complexity}\label{sec:complexity}
1766 The complexity of how many text selections that need to be analyzed for a body
1767 of in total $n$ statements, is bounded by $O(n^2)$. A body of statements is here
1768 all the statements in all nesting levels of a sequence of statements. A method
1769 body (or a block) is a body of statements. To prove that the complexity is
1770 bounded by $O(n^2)$, I present a couple of theorems and prove them.
1773 The number of text selections that need to be analyzed for each list of
1774 statements of length $n$, is exactly
1777 \sum_{i=1}^{n} i = \frac{n(n+1)}{2}
1778 \label{eq:complexityStatementList}
1780 \label{thm:numberOfTextSelection}
1784 For $n=1$ this is trivial: $\frac{1(1+1)}{2} = \frac{2}{2} = 1$. One statement
1785 equals one selection.
1787 For $n=2$, you get one text selection for the first statement, one selection
1788 for the second statement, and one selection for the two of them combined.
1789 This equals three selections. $\frac{2(2+1)}{2} = \frac{6}{2} = 3$.
1791 For $n=3$, you get 3 selections for the two first statements, as in the case
1792 where $n=2$. In addition you get one selection for the third statement itself,
1793 and two more statements for the combinations of it with the two previous
1794 statements. This equals six selections. $\frac{3(3+1)}{2} = \frac{12}{2} = 6$.
1796 Assume that for $n=k$ there exists $\frac{k(k+1)}{2}$ text selections. Then we
1797 want to add selections for another statement, following the previous $k$
1798 statements. So, for $n=k+1$, we get one additional selection for the statement
1799 itself. Then we get one selection for each pair of the new selection and the
1800 previous $k$ statements. So the total number of selections will be the number
1801 of already generated selections, plus $k$ for every pair, plus one for the
1802 statement itself: $\frac{k(k+1)}{2} + k +
1803 1 = \frac{k(k+1)+2k+2}{2} = \frac{k(k+1)+2(k+1)}{2} = \frac{(k+1)(k+2)}{2} =
1804 \frac{(k+1)((k+1)+1)}{2} = \sum_{i=1}^{k+1} i$
1807 %\definition{A \emph{body of statements} is a sequence of statements where every
1808 %statement may have sub-statements.}
1811 The number of text selections for a body of statements is maximized if all the
1812 statements are at the same level.
1813 \label{thm:textSelectionsMaximized}
1817 Assume we have a body of, in total, $k$ statements. Then, the sum of the
1818 lengths of all the lists of statements in the body, is also $k$. Let
1819 $\{l,\ldots,m,(k-l-\ldots-m)\}$ be the lengths of the lists of statements in
1820 the body, with $l+\ldots+m<k \Rightarrow \forall i \in \{l,\ldots,m\} : i < k$.
1822 Then, the number of text selections that are generated for the $k$ statements
1828 \frac{l(l+1)}{2} + \ldots + \frac{m(m+1)}{2} +
1829 \frac{(k-l-\ldots-m)((k-l-\ldots-m)+ 1)}{2} = \\
1830 \frac{l^2+l}{2} + \ldots + \frac{m^2+m}{2} + \frac{k^2 - 2kl - \ldots - 2km +
1831 l^2 + \ldots + m^2 + k - l - \ldots - m}{2} = \\
1832 \frac{2l^2 - 2kl + \ldots + 2m^2 - 2km + k^2 + k}{2}
1836 \noindent It then remains to show that this inequality holds:
1839 \frac{2l^2 - 2kl + \ldots + 2m^2 - 2km + k^2 + k}{2} < \frac{k(k+1)}{2} =
1843 \noindent By multiplication by $2$ on both sides, and by removing the equal
1847 2l^2 - 2kl + \ldots + 2m^2 - 2km < 0
1850 Since $\forall i \in \{l,\ldots,m\} : i < k$, we have that $\forall i \in
1851 \{l,\ldots,m\} : 2ki > 2i^2$, so all the pairs of parts on the form $2i^2-2ki$
1852 are negative. In sum, the inequality holds.
1856 Therefore, the complexity for the number of selections that need to be analyzed
1857 for a body of $n$ statements is $O\bigl(\frac{n(n+1)}{2}\bigr) = O(n^2)$.
1859 \section{Disqualifying a selection}
1860 Certain text selections would lead to broken code if used as input to the
1861 \ExtractAndMoveMethod refactoring. To avoid this, we have to check all text
1862 selections for such conditions before they are further analyzed. This section
1863 is therefore going to present some properties that make a selection unsuitable
1864 for our refactoring. When analyzing all these properties, it is assumed that the
1865 source code does not contain any compilation errors.
1867 \subsection{A call to a protected or package-private method}
1868 If a text selection contains a call to a protected or package-private method, it
1869 would not be safe to move it to another class. The reason for this, is that we
1870 cannot know if the called method is being overridden by some subclass of the
1871 \gloss{enclosingClass}, or not.
1873 Imagine that the protected method \method{foo} is declared in class \m{A},
1874 and overridden in class \m{B}. The method \method{foo} is called from within a
1875 selection done to a method in \m{A}. We want to extract and move this selection
1876 to another class. The method \method{foo} is not public, so the \MoveMethod
1877 refactoring must make it public, making the extracted method able to call it
1878 from the extracted method's new location. The problem is, that the now public
1879 method \method{foo} is overridden in a subclass, where it has a protected
1880 status. This makes the compiler complain that the subclass \m{B} is trying to
1881 reduce the visibility of a method declared in its superclass \m{A}. This is not
1882 allowed in Java, and for good reasons. It would make it possible to make a
1883 subclass that could not be a substitute for its superclass.
1885 The problem this check helps to avoid, is a little subtle. The problem does not
1886 arise in the class where the change is done, but in a class derived from it.
1887 This shows that classes acting as superclasses are especially fragile to
1888 introducing errors in the context of automated refactoring.
1890 This is also shown in bug\ldots \todoin{File Eclipse bug report}
1893 \subsection{A double class instance creation}
1894 The following is a problem caused solely by the underlying \MoveMethod
1895 refactoring. The problem occurs if two classes are instantiated such that the
1896 first constructor invocation is an argument to a second, and that the first
1897 constructor invocation takes an argument that is built up using a field. As an
1898 example, say that \var{name} is a field of the enclosing class, and we have the
1899 expression \code{new A(new B(name))}. If this expression is located in a
1900 selection that is moved to another class, \var{name} will be left untouched,
1901 instead of being prefixed with a variable of the same type as it is declared in.
1902 If \var{name} is the destination for the move, it is not replaced by
1903 \code{this}, or removed if it is a prefix to a member access
1904 (\code{name.member}), but it is still left by itself.
1906 Situations like this would lead to code that will not compile. Therefore, we
1907 have to avoid them by not allowing selections to contain such double class
1908 instance creations that also contain references to fields.
1910 \todoin{File Eclipse bug report}
1913 \subsection{Instantiation of non-static inner class}
1914 When a non-static inner class is instantiated, this must happen in the scope of
1915 its declaring class. This is because it must have access to the members of the
1916 declaring class. If the inner class is public, it is possible to instantiate it
1917 through an instance of its declaring class, but this is not handled by the
1918 underlying \MoveMethod refactoring.
1920 Performing a move on a method that instantiates a non-static inner class, will
1921 break the code if the instantiation is not handled properly. For this reason,
1922 selections that contain instantiations of non-static inner classes are deemed
1923 unsuitable for the \ExtractAndMoveMethod refactoring.
1925 \subsection{References to enclosing instances of the enclosing class}
1926 To ``reference an enclosing instance of the enclosing class'' is to reference
1927 another instance than the one for the immediately enclosing class. Imagine there
1928 is a (non-static) class \m{C} that is declared in the inner scope of another
1929 class. That class can again be nested inside a third class, and so on. Hence,
1930 the nested class \m{C} can have access to many enclosing instances of its
1931 innermost enclosing class. A selection in a method declared in class \m{C} is
1932 disqualified if it contains a statement that contains a reference to one or more
1933 instances of these enclosing classes of \m{C}.
1935 The problem with this, is that these references may not be valid if they are
1936 moved to another class. Theoretically, some situations could easily be solved by
1937 passing, to the moved method, a reference to the instance where the problematic
1938 referenced member is declared. This should work in the case where this member is
1939 publicly accessible. This is not done in the underlying \MoveMethod refactoring,
1940 so it cannot be allowed in the \ExtractAndMoveMethod refactoring either.
1942 \subsection{Inconsistent return statements}
1943 To verify that a text selection is consistent with respect to return statements,
1944 we must check that if a selection contains a return statement, then every
1945 possible execution path within the selection ends in either a return or a throw
1946 statement. This property is important regarding the \ExtractMethod refactoring.
1947 If it holds, it means that a method could be extracted from the selection, and a
1948 call to it could be substituted for the selection. If the method has a non-void
1949 return type, then a call to it would also be a valid return point for the
1950 calling method. If its return value is of the void type, then the \ExtractMethod
1951 refactoring will append an empty return statement to the back of the method
1952 call. Therefore, the analysis does not discriminate on either kind of return
1953 statements, with or without a return value.
1955 A \emph{throw} statement is accepted anywhere a return statement is required.
1956 This is because a throw statement causes an immediate exit from the current
1957 block, together with all outer blocks in its control flow that does not catch
1958 the thrown exception.
1960 We separate between explicit and implicit return statements. An \emph{explicit}
1961 return statement is formed by using the \code{return} keyword, while an
1962 \emph{implicit} return statement is a statement that is not formed using
1963 \code{return}, but must be the last statement of a method that can have any side
1964 effects. This can happen in methods with a void return type. An example is a
1965 statement that is inside one or more blocks. The last statement of a method
1966 could for instance be a synchronized statement, but the last statement that is
1967 executed in the method, and that can have any side effects, may be located
1968 inside the body of the synchronized statement.
1970 We can start the check for this property by looking at the last statement of a
1971 selection to see if it is a return statement (explicit or implicit) or a throw
1972 statement. If this is the case, then the property holds, assuming the selected
1973 code do not contain any compilation errors. All execution paths within the
1974 selection should end in either this, or another, return or throw statement.
1976 If the last statement of the selection is not a \emph{return} or \emph{throw},
1977 the execution of it must eventually end in one of these types of statements for
1978 the selection to be legal. This means that all branches of the last statement of
1979 every branch must end in a return or throw. Given this recursive definition,
1980 there are only five types of statements that are guaranteed to end in a return
1981 or throw if their child branches do. All other statements would have to be
1982 considered illegal. The first three: Block-statements, labeled statements and
1983 do-statements are all kinds of fall-through statements that always get their
1984 body executed. Do-statements would not make much sense if written such that they
1985 always end after the first round of execution of their body, but that is not our
1986 concern. The remaining two statements that can end in a return or throw are
1987 if-statements and try-statements.
1989 For an if-statement, the rule is that if its then-part does not contain any
1990 return or throw statements, this is considered illegal. If the then-part does
1991 contain a return or throw, the else-part is checked. If its else-part is
1992 non-existent, or it does not contain any return or throw statements, the
1993 statement is considered illegal. If an if-statement is not considered illegal,
1994 the bodies of its two parts must be checked.
1996 Try-statements are handled much the same way as if-statements. The body of a
1997 try-statement must contain a return or throw. The same applies to its catch
1998 clauses and finally body. \todoin{finally body?}
2000 \subsection{Ambiguous return values}
2001 The problem with ambiguous return values arises when a selection is chosen to be
2002 extracted into a new method, but if refactored it needs to return more than one
2003 value from that method.
2005 This problem occurs in two situations. The first situation arises when there is
2006 more than one local variable that is both assigned to within a selection and
2007 also referenced after the selection. The other situation occurs when there is
2008 only one such assignment, but the selection also contain return statements.
2010 Therefore we must examine the selection for assignments to local variables that
2011 are referenced after the text selection. Then we must verify that not more than
2012 one such reference is done, or zero if any return statements are found.
2014 \subsection{Illegal statements}
2015 An illegal statement may be a statement that is of a type that is never allowed,
2016 or it may be a statement of a type that is only allowed if certain conditions
2019 Any use of the \var{super} keyword is prohibited, since its meaning is altered
2020 when moving a method to another class.
2022 For a \emph{break} statement, there are two situations to consider: A break
2023 statement with or without a label. If the break statement has a label, it is
2024 checked that whole of the labeled statement is inside the selection. If the
2025 break statement does not have a label attached to it, it is checked that its
2026 innermost enclosing loop or switch statement also is inside the selection.
2028 The situation for a \emph{continue} statement is the same as for a break
2029 statement, except that it is not allowed inside switch statements.
2031 Regarding \emph{assignments}, two types of assignments are allowed: Assignments
2032 to non-final variables and assignments to array access. All other assignments
2033 are regarded illegal.
2035 \paragraph{Incompleteness.} The list of illegal statements is not complete, and
2036 a lot of situations that can lead to compilation errors or behavior changes are
2037 not considered. It is not feasible to consider all such situations within the
2038 limits of this master's project, and maybe not outside of them either. The
2039 feasibility of this problem could be explored further by others.
2041 \section{Disqualifying selections from the
2042 example}\label{sec:disqualifyingExample}
2043 Among the selections we found for the code in \myref{lst:grandExample}, not many
2044 of them must be disqualified on the basis of containing something illegal. The
2045 only statement causing trouble is the break statement in line 18. None of the
2046 selections on nesting level 2 can contain this break statement, since the
2047 innermost switch statement is not inside any of these selections.
2049 This means that the text selections $(18)$, $(16,18)$ and $(17,18)$ can be
2050 excluded from further consideration, and we are left with the following
2054 \item[Level 1 (10 selections)] \hfill \\
2055 $\{(5,9), (11), (12), (14,21), (5,11), (5,12), (5,21), (11,12),
2056 (11,21), \\(12,21)\}$
2058 \item[Level 2 (10 selections)] \hfill \\
2059 $\{(6), (7), (8), (6,7), (6,8), (7,8), (16), (17), (16,17), (20)\}$
2062 \section{Finding a move target}
2063 In the analysis needed to perform the \ExtractAndMoveMethod refactoring
2064 automatically, the selection we choose is found among all the selections that
2065 have a possible move target. Therefore, the best possible move target must be
2066 found for all the candidate selections, so that we are able to sort out the
2067 selection that is best suited for the refactoring.
2069 To find the best move target for a specific text selection, we first need to
2070 find all the possible targets. Since the target must be a local variable or a
2071 field, we are basically looking for names within the selection; names that
2072 represents references to variables.
2074 The names we are looking for, we call prefixes. This is because we are not
2075 interested in names that occur in the middle of a dot-separated sequence of
2076 names. We are only interested in names constituting prefixes of other names, and
2077 possibly themselves. The reason for this, is that two lexically equal names need
2078 not be referencing the same variable, if they themselves are not referenced via
2079 the same prefix. Consider the two method calls \code{a.x.foo()} and
2080 \code{b.x.foo()}. Here, the two references to \code{x}, in the middle of the
2081 qualified names both preceding \code{foo()}, are not referencing the same
2082 variable. Even though the variables may share the type, and the method
2083 \method{foo} thus is the same for both, we would not know through which of the
2084 variables \var{a} or \var{b} we should call the extracted method.
2086 The possible move targets are then the prefixes that are not among a subset of
2087 the prefixes that are not valid move targets \see{s:unfixes}. Also, prefixes
2088 that are just simple names, and have only one occurrence, are left out. This is
2089 because they are not going to have any positive effect on coupling between
2090 classes, and are only going to increase the complexity of the code.
2092 For finding the best move target among these safe prefixes, a simple heuristic
2093 is used. It is as simple as choosing the prefix that is most frequently
2094 referenced within the selection.
2096 \section{Unfixes}\label{s:unfixes}
2097 We will call the prefixes that are not valid as move targets for unfixes.
2099 A name that is assigned to within a selection can be an unfix. The reason for
2100 this is that the result would be an assignment to the \type{this} keyword, which
2101 is not valid in Java \see{eclipse_bug_420726}.
2103 Prefixes that originate from variable declarations within the same selection are
2104 also considered unfixes. The reason for this is that when a method is moved, it
2105 needs to be called through a variable. If this variable is also declared within
2106 the method that is to be moved, this obviously cannot be done.
2108 Also considered as unfixes are variable references that are of types that are
2109 not suitable for moving methods to. This can either be because it is not
2110 physically possible to move a method to the desired class or that it will cause
2111 compilation errors by doing so.
2113 If the type binding for a name is not resolved it is considered an unfix. The
2114 same applies to types that are only found in compiled code, so they have no
2115 underlying source that is accessible to us. (E.g. the \type{java.lang.String}
2118 Interface types are not suitable as targets. This is simply because interfaces
2119 in Java cannot contain methods with bodies. (This thesis does not deal with
2120 features of Java versions later than Java 7. Java 8 has interfaces with default
2121 implementations of methods.)
2123 Neither are local types allowed. This accounts for both local and anonymous
2124 classes. Anonymous classes are effectively the same as interface types with
2125 respect to unfixes. Local classes could in theory be used as targets, but this
2126 is not possible due to limitations of the way the \refa{Extract and Move Method}
2127 refactoring has to be implemented. The problem is that the refactoring is done
2128 in two steps, so the intermediate state between the two refactorings would not
2129 be legal Java code. In the intermediate step for the case where a local class is
2130 the move target, the extracted method would need to take the local class as a
2131 parameter. This new method would need to live in the scope of the declaring
2132 class of the originating method. The local class would then not be in the scope
2133 of the extracted method, thus bringing the source code into an illegal state.
2134 This scenario is shown in \myref{lst:extractMethodLocalClass}. One could imagine
2135 that the method was extracted and moved in one operation, without an
2136 intermediate state. Then it would make sense to include variables with types of
2137 local classes in the set of legal targets, since the local classes would then be
2138 in the scopes of the method calls. If this makes any difference for software
2139 metrics that measure coupling would be a different discussion.
2141 \todoin{highlight code!}
2143 \begin{listing}[htb]
2144 \begin{multicols}{2}
2145 \begin{minted}[frame=topline,label=Before,framesep=\mintedframesep]{java}
2146 void declaresLocalClass() {
2161 \begin{minted}[frame=topline,label={After Extract
2162 Method},framesep=\mintedframesep]{java}
2163 void declaresLocalClass() {
2174 // Illegal intermediate step
2175 void fooBar(LocalClass inst) {
2181 \caption{The \refa{Extract and Move Method} refactoring bringing the code into
2182 an illegal state with an intermediate step.}
2183 \label{lst:extractMethodLocalClass}
2186 The last class of names that are considered unfixes are names used in null
2187 tests. These are tests that read like this: if \code{<name>} equals \var{null}
2188 then do something. If allowing variables used in those kinds of expressions as
2189 targets for moving methods, we would end up with code containing boolean
2190 expressions like \code{this == null}, which would always evaluate to
2191 \code{false}, since \var{this} would never be \var{null}. The existence of a
2192 null test indicates that a variable is expected to sometimes hold the value
2193 \var{null}. By using a variable used in a null test as a move target, we could
2194 potentially end up with a
2195 null pointer exception if the method is called on a variable with a null value.
2197 \section{Finding the example selections that have possible targets}
2198 We now pick up the thread from \myref{sec:disqualifyingExample} where we have a
2199 set of text selections that need to be analyzed to find out if some of them are
2200 suitable targets for the \ExtractAndMoveMethod refactoring.
2202 We start by analyzing the text selections for nesting level 2, because these
2203 results can be used to reason about the selections for nesting level 1. First we
2204 have all the single-statement selections.
2207 \item[Selections $(6)$, $(8)$ and $(20)$.] \hfill \\
2208 All these selections have a prefix that contains a possible target, namely
2209 the variable \var{a}. The problem is that the prefixes are only one segment
2210 long, and their frequency counts are only 1 as well. None of these
2211 selections are therefore considered as suitable candidates for the
2214 \item[Selection $(7)$.] \hfill \\
2215 This selection contains the unfix \var{a}, and no other possible targets.
2216 The reason for \var{a} being an unfix is that it is assigned to within the
2217 selection. Selection $(7)$ is therefore unsuited as a refactoring candidate.
2219 \item[Selections $(16)$ and $(17)$.] \hfill \\
2220 These selections both have a possible target. The target for both selections
2221 is the variable \var{b}. Both the prefixes have frequency 1. We denote this
2222 with the new tuples $((16), \texttt{b.a}, f(1))$ and $((17), \texttt{b.a},
2223 f(1))$. They contain the selection, the prefix with the target and the
2224 frequency for this prefix.
2228 Then we have all the text selections from level 2 that are composed of multiple
2232 \item[Selections $(6,7)$, $(6,8)$ and $(7,8)$.] \hfill \\
2233 All these selections are disqualified for the reason that they contain the
2234 unfix \var{a}, due to the assignment, and no other possible move targets.
2236 \item[Selection $(16,17)$.] \hfill \\
2237 This selection is the last selection that is not analyzed on nesting level
2238 2. It contains only one possible move target, and that is the variable
2239 \var{b}. It also contains only one prefix \var{b.a}, with frequency count
2240 2. Therefore we have a new candidate $((16,17), \texttt{b.a}, f(2))$.
2244 Moving on to the text selections for nesting level 1, starting with the
2245 single-statement selections:
2248 \item[Selection $(5,9)$.] \hfill \\
2249 This selection contains two variable references that must be analyzed to see
2250 if they are possible move candidates. The first one is the variable
2251 \var{bool}. This variable is of type \type{boolean}, which is a primary type
2252 and therefore not possible to make any changes to. The second variable is
2253 \var{a}. The variable \var{a} is an unfix in $(5,9)$, for the same reason as
2254 in the selections $(6,7)$, $(7,8)$ and $(6,8)$. So selection $(5,9)$
2255 contains no possible move targets.
2257 \item[Selections $(11)$ and $(12)$.] \hfill \\
2258 These selections are disqualified for the same reasons as selections $(6)$
2259 and $(8)$. Their prefixes are one segment long and are referenced only one
2262 \item[Selection $(14,21)$] \hfill \\
2263 This is the switch statement from \myref{lst:grandExample}. It contains the
2264 relevant variable references \var{val}, \var{a} and \var{b}. The variable
2265 \var{val} is a primary type, just as \var{bool}. The variable \var{a} is
2266 only found in one statement, and in a prefix with only one segment, so it is
2267 not considered to be a possible move target. The only variable left is
2268 \var{b}. Just as in the selection $(16,17)$, \var{b} is part of the prefix
2269 \code{b.a}, which has 2 appearances. We have found a new candidate
2270 $((14,21), \texttt{b.a}, f(2))$.
2274 It remains to see if we get any new candidates by analyzing the multi-statement
2275 text selections for nesting level 1:
2278 \item[Selections $(5,11)$ and $(5,12)$.] \hfill \\
2279 These selections are disqualified for the same reason as $(5,9)$. The only
2280 possible move target \var{a} is an unfix.
2282 \item[Selection $(5,21)$.] \hfill \\
2283 This is whole of the method body in \myref{lst:grandExample}. The variables
2284 \var{a}, \var{bool} and \var{val} are either an unfix or primary types. The
2285 variable \var{b} is the only possible move target, and we have, again, the
2286 prefix \code{b.a} as the center of attention. The refactoring candidate is
2287 $((5,21), \texttt{b.a}, f(2))$.
2289 \item[Selection $(11,12)$.] \hfill \\
2290 This small selection contains the prefix \code{a} with frequency 2, and no
2291 unfixes. The candidate is $((11,12), \texttt{a}, f(2))$.
2293 \item[Selection $(11,21)$] \hfill \\
2294 This selection contains the selection $(11,12)$ in addition to the switch
2295 statement. The selection has two possible move targets. The first one is
2296 \var{b}, in a prefix with frequency 2. The second is \var{a}, although it
2297 is in a simple prefix, it is referenced 3 times, and is therefore chosen
2298 as the target for the selection. The new candidate is $((11,21),
2301 \item[Selection $(12,21)$.] \hfill \\
2302 This selection is similar to the previous $(11,21)$, only that \var{a} now
2303 has frequency count 2. This means that the prefixes \code{a} and
2304 \code{b.a} both have the count 2. Of the two, \code{b.a} is preferred,
2305 since it has more segments than \code{a}. Thus the candidate for this
2306 selection is $((12,21), \texttt{b.a}, f(2))$.
2310 For the method in \myref{lst:grandExample} we therefore have the following 8
2311 candidates for the \ExtractAndMoveMethod refactoring: $\{((16), \texttt{b.a},
2312 f(1)), ((17), \texttt{b.a}, f(1)), ((16,17), \texttt{b.a}, f(2)), ((14,21),
2313 \texttt{b.a}, f(2)), \\ ((5,21), \texttt{b.a}, f(2)), ((11,12), \texttt{a},
2314 f(2)), ((11,21), \texttt{a}, f(3)), ((12,21), \texttt{b.a}, f(2))\}$.
2316 It now only remains to specify an order for these candidates, so we can choose
2317 the most suitable one to refactor.
2320 \section{Choosing the selection}\label{sec:choosingSelection}
2321 When choosing a selection between the text selections that have possible move
2322 targets, the selections need to be ordered. The criteria below are presented in
2323 the order they are prioritized. If not one selection is favored over the other
2324 for a concrete criterion, the selections are evaluated by the next criterion.
2327 \item The first criterion that is evaluated is whether a selection contains
2328 any unfixes or not. If selection \m{A} contains no unfixes, while selection
2329 \m{B} does, selection \m{A} is favored over selection \m{B}. This is
2330 because, if we can, we want to avoid moving such as assignments and variable
2331 declarations. This is done under the assumption that, if possible, avoiding
2332 selections containing unfixes will make the code moved a little cleaner.
2334 \item The second criterion that is evaluated is whether a selection contains
2335 multiple possible targets or not. If selection \m{A} has only one possible
2336 target, and selection \m{B} has multiple, selection \m{A} is favored. If
2337 both selections have multiple possible targets, they are considered equal
2338 with respect to this criterion. The rationale for this heuristic is that we
2339 would prefer not to introduce new couplings between classes when performing
2340 the \ExtractAndMoveMethod refactoring.
2342 \item When evaluating this criterion, this is with the knowledge that
2343 selection \m{A} and \m{B} both have one possible target, or multiple
2344 possible targets. Then, if the move target candidate of selection \m{A} has
2345 a higher reference count than the target candidate of selection \m{B},
2346 selection \m{A} is favored. The reason for this is that we would like to
2347 move the selection that gets rid of the most references to another class.
2349 \item The last criterion is that if the frequencies of the targets chosen for
2350 both selections are equal, the selection with the target that is part of the
2351 prefix with highest number of segments is favored. This is done to favor
2356 If none of the above mentioned criteria favors one selection over another, the
2357 selections are considered to be equally good candidates for the
2358 \ExtractAndMoveMethod refactoring.
2360 \section{Performing changes}
2361 When a text selection and a move target is found for the \ExtractAndMoveMethod
2362 refactoring, the actual changes are executed by two existing primitive
2363 refactorings. First the \ExtractMethod refactoring is used to extract the
2364 selection into a new method. Then the \MoveMethod refactoring is used to move
2365 that new method to the class determined by the move target.
2367 If, at any point, an exception is thrown or the preconditions for one of the
2368 primitive refactorings are not satisfied, the composite refactoring is aborted,
2369 and the source code is left in its current state. This has the implication that
2370 the \ExtractAndMoveMethod refactoring could end up being partially executed.
2371 This happens if the \ExtractMethod refactoring is executed, but the \MoveMethod
2372 refactoring is being canceled. A partial execution is not considered a problem,
2373 since the code should still compile.
2375 \todoin{Pointing to implementation chapter}
2377 \section{Concluding the example}
2378 For choosing one of the remaining selections, we need to order our candidates
2379 after the criteria in the previous section. Below we have the candidates ordered
2380 in descending order, with the ``best'' candidate first:
2382 \begin{multicols}{2}
2384 \item $((16,17), \texttt{b.a}, f(2))$
2385 \item $((11,12), \texttt{a}, f(2))$
2386 \item $((16), \texttt{b.a}, f(1))$
2387 \item $((17), \texttt{b.a}, f(1))$
2390 % Many possible targets
2391 \item $((11,21), \texttt{a}, f(3))$
2392 \item $((5,21), \texttt{b.a}, f(2))$
2393 \item $((12,21), \texttt{b.a}, f(2))$
2394 \item $((14,21), \texttt{b.a}, f(2))$
2419 The candidates ordered 5-8 all have unfixes in them, therefore they are ordered
2420 last. None of the candidates ordered 1-4 have multiple possible targets. The
2421 only interesting discussion is now why $(16,17)$ was favored over $(11,12)$.
2422 This is because the prefix \code{b.a} enclosing the move target of selection
2423 $(16,17)$ has one more segment than the prefix \code{a} of $(11,12)$.
2425 The selection is now extracted into a new method \method{gen\_123} and then
2426 moved to class \type{B}, since that is the type of the variable \var{b} that is
2427 the target from the chosen refactoring candidate. The name of the method has a
2428 randomly generated suffix. In the refactoring implementation, the extracted
2429 methods follow the pattern \code{generated\_<long>}, where \code{<long>} is a
2430 pseudo-random long value. This is shortened here to make the example readable.
2431 The result after the refactoring is shown in \myref{lst:grandExampleResult}.
2433 \begin{listing}[htb]
2434 \begin{multicols}{2}
2435 \begin{minted}[linenos]{java}
2437 A a; B b; boolean bool;
2439 void method(int val) {
2462 \begin{minted}[]{java}
2466 public void gen_123(C c) {
2474 \caption{The result after refactoring the class \type{C} of
2475 \myref{lst:grandExample} with the \ExtractAndMoveMethod refactoring with
2476 $((16,17), \texttt{b.a}, f(2))$ as input.}
2477 \label{lst:grandExampleResult}
2481 \chapter{Refactorings in Eclipse JDT: Design and
2482 shortcomings}\label{ch:jdt_refactorings}
2484 This chapter will deal with some of the design behind refactoring support in
2485 \name{Eclipse}, and the JDT in specific. After which it will follow a section
2486 about shortcomings of the refactoring API in terms of composition of
2490 The refactoring world of \name{Eclipse} can in general be separated into two parts: The
2491 language independent part and the part written for a specific programming
2492 language -- the language that is the target of the supported refactorings.
2493 \todo{What about the language specific part?}
2495 \subsection{The Language Toolkit}
2496 The Language Toolkit\footnote{The content of this section is a mixture of
2497 written material from
2498 \url{https://www.eclipse.org/articles/Article-LTK/ltk.html} and
2499 \url{http://www.eclipse.org/articles/article.php?file=Article-Unleashing-the-Power-of-Refactoring/index.html},
2500 the LTK source code and my own memory.}, or LTK for short, is the framework that
2501 is used to implement refactorings in \name{Eclipse}. It is language independent and
2502 provides the abstractions of a refactoring and the change it generates, in the
2503 form of the classes \typewithref{org.eclipse.ltk.core.refactoring}{Refactoring}
2504 and \typewithref{org.eclipse.ltk.core.refactoring}{Change}.
2506 There are also parts of the LTK that is concerned with user interaction, but
2507 they will not be discussed here, since they are of little value to us and our
2508 use of the framework. We are primarily interested in the parts that can be
2511 \subsubsection{The Refactoring Class}
2512 The abstract class \type{Refactoring} is the core of the LTK framework. Every
2513 refactoring that is going to be supported by the LTK has to end up creating an
2514 instance of one of its subclasses. The main responsibilities of subclasses of
2515 \type{Refactoring} are to implement template methods for condition checking
2516 (\methodwithref{org.eclipse.ltk.core.refactoring.Refactoring}{checkInitialConditions}
2518 \methodwithref{org.eclipse.ltk.core.refactoring.Refactoring}{checkFinalConditions}),
2520 \methodwithref{org.eclipse.ltk.core.refactoring.Refactoring}{createChange}
2521 method that creates and returns an instance of the \type{Change} class.
2523 If the refactoring shall support that others participate in it when it is
2524 executed, the refactoring has to be a processor-based
2525 refactoring\typeref{org.eclipse.ltk.core.refactoring.participants.ProcessorBasedRefactoring}.
2526 It then delegates to its given
2527 \typewithref{org.eclipse.ltk.core.refactoring.participants}{RefactoringProcessor}
2528 for condition checking and change creation. Participating in a refactoring can
2529 be useful in cases where the changes done to programming source code affect
2530 other related resources in the workspace. This can be names or paths in
2531 configuration files, or maybe one would like to perform additional logging of
2532 changes done in the workspace.
2534 \subsubsection{The Change Class}
2535 This class is the base class for objects that is responsible for performing the
2536 actual workspace transformations in a refactoring. The main responsibilities for
2537 its subclasses are to implement the
2538 \methodwithref{org.eclipse.ltk.core.refactoring.Change}{perform} and
2539 \methodwithref{org.eclipse.ltk.core.refactoring.Change}{isValid} methods. The
2540 \method{isValid} method verifies that the change object is valid and thus can be
2541 executed by calling its \method{perform} method. The \method{perform} method
2542 performs the desired change and returns an undo change that can be executed to
2543 reverse the effect of the transformation done by its originating change object.
2545 \subsubsection{Executing a Refactoring}\label{executing_refactoring}
2546 The life cycle of a refactoring generally follows two steps after creation:
2547 condition checking and change creation. By letting the refactoring object be
2549 \typewithref{org.eclipse.ltk.core.refactoring}{CheckConditionsOperation} that
2550 in turn is handled by a
2551 \typewithref{org.eclipse.ltk.core.refactoring}{CreateChangeOperation}, it is
2552 assured that the change creation process is managed in a proper manner.
2554 The actual execution of a change object has to follow a detailed life cycle.
2555 This life cycle is honored if the \type{CreateChangeOperation} is handled by a
2556 \typewithref{org.eclipse.ltk.core.refactoring}{PerformChangeOperation}. If also
2557 an undo manager\typeref{org.eclipse.ltk.core.refactoring.IUndoManager} is set
2558 for the \type{PerformChangeOperation}, the undo change is added into the undo
2561 \section{Shortcomings}
2562 This section is introduced naturally with a conclusion: The JDT refactoring
2563 implementation does not facilitate composition of refactorings.
2564 \todo{refine}This section will try to explain why, and also identify other
2565 shortcomings of both the usability and the readability of the JDT refactoring
2568 I will begin at the end and work my way toward the composition part of this
2571 \subsection{Absence of generics in Eclipse source code}
2572 This section is not only concerning the JDT refactoring API, but also large
2573 quantities of the \name{Eclipse} source code. The code shows a striking absence of the
2574 Java language feature of generics. It is hard to read a class' interface when
2575 methods return objects or takes parameters of raw types such as \type{List} or
2576 \type{Map}. This sometimes results in having to read a lot of source code to
2577 understand what is going on, instead of relying on the available interfaces. In
2578 addition, it results in a lot of ugly code, making the use of typecasting more
2579 of a rule than an exception.
2581 \subsection{Composite refactorings will not appear as atomic actions}
2582 When composing primitive refactorings from the JDT, it is not possible to make
2583 them appear as being executed as one change, but only as multiple small changes.
2585 \subsubsection{Missing Flexibility from JDT Refactorings}
2586 The JDT refactorings are not made with composition of refactorings in mind. When
2587 a JDT refactoring is executed, it assumes that all conditions for it to be
2588 applied successfully can be found by reading source files that have been
2589 persisted to disk. They can only operate on the actual source material, and not
2590 (in-memory) copies thereof. This constitutes a major disadvantage when trying to
2591 compose refactorings, since if an exception occurs in the middle of a sequence
2592 of refactorings, it can leave the project in a state where the composite
2593 refactoring was only partially executed. It makes it hard to discard the changes
2594 done without monitoring and consulting the undo manager, an approach that is not
2597 \subsubsection{Broken Undo History}
2598 When designing a composed refactoring that is to be performed as a sequence of
2599 refactorings, you would like it to appear as a single change to the workspace.
2600 This implies that you would also like to be able to undo all the changes done by
2601 the refactoring in a single step. This is not the way it appears when a sequence
2602 of JDT refactorings is executed. It leaves the undo history filled up with
2603 individual undo actions corresponding to every single JDT refactoring in the
2604 sequence. This problem is not trivial to handle in \name{Eclipse}
2605 \see{hacking_undo_history}.
2609 \chapter{Composite refactorings in Eclipse}
2611 \section{A simple ad hoc model}
2612 As pointed out in \myref{ch:jdt_refactorings}, the \name{Eclipse} JDT refactoring model
2613 is not very well suited for making composite refactorings. Therefore a simple
2614 model using changer objects (of type \type{RefaktorChanger}) is used as an
2615 abstraction layer on top of the existing \name{Eclipse} refactorings, instead of
2616 extending the \typewithref{org.eclipse.ltk.core.refactoring}{Refactoring} class.
2618 The use of an additional abstraction layer is a deliberate choice. It is due to
2619 the problem of creating a composite
2620 \typewithref{org.eclipse.ltk.core.refactoring}{Change} that can handle text
2621 changes that interfere with each other. Thus, a \type{RefaktorChanger} may, or
2622 may not, take advantage of one or more existing refactorings, but it is always
2623 intended to make a change to the workspace.
2625 \subsection{A typical \type{RefaktorChanger}}
2626 The typical refaktor changer class has two responsibilities, checking
2627 preconditions and executing the requested changes. This is not too different
2628 from the responsibilities of an LTK refactoring, with the distinction that a
2629 refaktor changer also executes the change, while an LTK refactoring is only
2630 responsible for creating the object that can later be used to do the job.
2632 Checking of preconditions is typically done by an
2633 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{Analyzer}. If the
2634 preconditions validate, the upcoming changes are executed by an
2635 \typewithref{no.uio.ifi.refaktor.change.executors}{Executor}.
2637 \section{The Extract and Move Method refactoring}
2638 %The Extract and Move Method Refactoring is implemented mainly using these
2641 % \item \type{ExtractAndMoveMethodChanger}
2642 % \item \type{ExtractAndMoveMethodPrefixesExtractor}
2643 % \item \type{Prefix}
2644 % \item \type{PrefixSet}
2647 \subsection{The building blocks}
2648 This is a composite refactoring, and hence is built up using several primitive
2649 refactorings. These basic building blocks are, as its name implies, the
2650 \ExtractMethod refactoring\citing{refactoring} and the \MoveMethod
2651 refactoring\citing{refactoring}. In \name{Eclipse}, the implementations of these
2652 refactorings are found in the classes
2653 \typewithref{org.eclipse.jdt.internal.corext.refactoring.code}{ExtractMethodRefactoring}
2655 \typewithref{org.eclipse.jdt.internal.corext.refactoring.structure}{MoveInstanceMethodProcessor},
2656 where the last class is designed to be used together with the processor-based
2657 \typewithref{org.eclipse.ltk.core.refactoring.participants}{MoveRefactoring}.
2659 \subsubsection{The ExtractMethodRefactoring class}
2660 This class is quite simple in its use. The only parameters it requires for
2661 construction is a compilation
2662 unit\typeref{org.eclipse.jdt.core.ICompilationUnit}, the offset into the source
2663 code where the extraction shall start, and the length of the source to be
2664 extracted. Then you have to set the method name for the new method together with
2665 its visibility and some not so interesting parameters.
2667 \subsubsection{The MoveInstanceMethodProcessor class}
2668 For the \refa{Move Method}, the processor requires a little more advanced input than
2669 the class for the \refa{Extract Method}. For construction it requires a method
2670 handle\typeref{org.eclipse.jdt.core.IMethod} for the method that is to be moved.
2671 Then the target for the move has to be supplied as the variable binding from a
2672 chosen variable declaration. In addition to this, some parameters have to be set
2673 regarding setters/getters, as well as delegation.
2675 To make the processor a working refactoring, a \type{MoveRefactoring} must be
2676 created with it as a parameter.
2678 \subsection{The ExtractAndMoveMethodChanger class}
2680 The \typewithref{no.uio.ifi.refaktor.changers}{ExtractAndMoveMethodChanger}
2681 class is a subclass of the class
2682 \typewithref{no.uio.ifi.refaktor.changers}{RefaktorChanger}. It is responsible
2683 for analyzing and finding the best target for, and also executing, a composition
2684 of the \refa{Extract Method} and \refa{Move Method} refactorings. This particular changer is
2685 the one of my changers that is closest to being a true LTK refactoring. It can
2686 be reworked to be one if the problems with overlapping changes are resolved. The
2687 changer requires a text selection and the name of the new method, or else a
2688 method name will be generated. The selection has to be of the type
2689 \typewithref{no.uio.ifi.refaktor.utils}{CompilationUnitTextSelection}. This
2690 class is a custom extension to
2691 \typewithref{org.eclipse.jface.text}{TextSelection}, that in addition to the
2692 basic offset, length and similar methods, also carry an instance of the
2693 underlying compilation unit handle for the selection.
2696 \type{ExtractAndMoveMethodAnalyzer}}\label{extractAndMoveMethodAnalyzer}
2697 The analysis and precondition checking is done by the
2698 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{ExtractAnd\-MoveMethodAnalyzer}.
2699 First is check whether the selection is a valid selection or not, with respect
2700 to statement boundaries and that it actually contains any selections. Then it
2701 checks the legality of both extracting the selection and also moving it to
2702 another class. This checking of is performed by a range of checkers
2703 \see{checkers}. If the selection is approved as legal, it is analyzed to find
2704 the presumably best target to move the extracted method to.
2706 For finding the best suitable target the analyzer is using a
2707 \typewithref{no.uio.ifi.refaktor.analyze.collectors}{PrefixesCollector} that
2708 collects all the possible candidate targets for the refactoring. All the
2709 non-candidates are found by an
2710 \typewithref{no.uio.ifi.refaktor.analyze.collectors}{UnfixesCollector} that
2711 collects all the targets that will give some kind of error if used. (For
2712 details about the property collectors, see \myref{propertyCollectors}.) All
2713 prefixes (and unfixes) are represented by a
2714 \typewithref{no.uio.ifi.refaktor.extractors}{Prefix}, and they are collected
2715 into sets of prefixes. The safe prefixes are found by subtracting from the set
2716 of candidate prefixes the prefixes that is enclosing any of the unfixes. A
2717 prefix is enclosing an unfix if the unfix is in the set of its sub-prefixes. As
2718 an example, \code{``a.b''} is enclosing \code{``a''}, as is \code{``a''}. The
2719 safe prefixes is unified in a \type{PrefixSet}. If a prefix has only one
2720 occurrence, and is a simple expression, it is considered unsuitable as a move
2721 target. This occurs in statements such as \code{``a.foo()''}. For such
2722 statements it bares no meaning to extract and move them. It only generates an
2723 extra method and the calling of it.
2725 The most suitable target for the refactoring is found by finding the prefix with
2726 the most occurrences. If two prefixes have the same occurrence count, but they
2727 differ in the number of segments, the one with most segments is chosen.
2730 \type{ExtractAndMoveMethodExecutor}}\label{extractAndMoveMethodExecutor}
2731 If the analysis finds a possible target for the composite refactoring, it is
2733 \typewithref{no.uio.ifi.refaktor.change.executors}{ExtractAndMoveMethodExecutor}.
2734 It is composed of the two executors known as
2735 \typewithref{no.uio.ifi.refaktor.change.executors}{ExtractMethodRefactoringExecutor}
2737 \typewithref{no.uio.ifi.refaktor.change.executors}{MoveMethodRefactoringExecutor}.
2738 The \type{ExtractAndMoveMethodExecutor} is responsible for gluing the two
2739 together by feeding the \type{MoveMethod\-RefactoringExecutor} with the
2740 resources needed after executing the extract method refactoring.
2742 \subsubsection{The \type{ExtractMethodRefactoringExecutor}}
2743 This executor is responsible for creating and executing an instance of the
2744 \type{ExtractMethodRefactoring} class. It is also responsible for collecting
2745 some post execution resources that can be used to find the method handle for the
2746 extracted method, as well as information about its parameters, including the
2747 variable they originated from.
2749 \subsubsection{The \type{MoveMethodRefactoringExecutor}}
2750 This executor is responsible for creating and executing an instance of the
2751 \type{MoveRefactoring}. The move refactoring is a processor-based refactoring,
2752 and for the \refa{Move Method} refactoring it is the \type{MoveInstanceMethodProcessor}
2755 The handle for the method to be moved is found on the basis of the information
2756 gathered after the execution of the \refa{Extract Method} refactoring. The only
2757 information the \type{ExtractMethodRefactoring} is sharing after its execution,
2758 regarding finding the method handle, is the textual representation of the new
2759 method signature. Therefore it must be parsed, the strings for types of the
2760 parameters must be found and translated to a form that can be used to look up
2761 the method handle from its type handle. They have to be on the unresolved form.
2762 The name for the type is found from the original selection, since an extracted
2763 method must end up in the same type as the originating method.
2765 When analyzing a selection prior to performing the \refa{Extract Method} refactoring, a
2766 target is chosen. It has to be a variable binding, so it is either a field or a
2767 local variable/parameter. If the target is a field, it can be used with the
2768 \type{MoveInstanceMethodProcessor} as it is, since the extracted method still is
2769 in its scope. But if the target is local to the originating method, the target
2770 that is to be used for the processor must be among its parameters. Thus the
2771 target must be found among the extracted method's parameters. This is done by
2772 finding the parameter information object that corresponds to the parameter that
2773 was declared on basis of the original target's variable when the method was
2774 extracted. (The extracted method must take one such parameter for each local
2775 variable that is declared outside the selection that is extracted.) To match the
2776 original target with the correct parameter information object, the key for the
2777 information object is compared to the key from the original target's binding.
2778 The source code must then be parsed to find the method declaration for the
2779 extracted method. The new target must be found by searching through the
2780 parameters of the declaration and choose the one that has the same type as the
2781 old binding from the parameter information object, as well as the same name that
2782 is provided by the parameter information object.
2786 SearchBasedExtractAndMoveMethodChanger}\label{searchBasedExtractAndMoveMethodChanger}
2788 \typewithref{no.uio.ifi.refaktor.change.changers}{SearchBasedExtractAndMoveMethodChanger}
2789 is a changer whose purpose is to automatically analyze a method, and execute the
2790 \ExtractAndMoveMethod refactoring on it if it is a suitable candidate for the
2793 First, the \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{SearchBasedExtractAndMoveMethodAnalyzer} is used
2794 to analyze the method. If the method is found to be a candidate, the result from
2795 the analysis is fed to the \type{ExtractAndMoveMethodExecutor}, whose job is to
2796 execute the refactoring \see{extractAndMoveMethodExecutor}.
2798 \subsubsection{The SearchBasedExtractAndMoveMethodAnalyzer}
2799 This analyzer is responsible for analyzing all the possible text selections of a
2800 method and then to choose the best result out of the analysis results that are,
2801 by the analyzer, considered to be the potential candidates for the
2802 \ExtractAndMoveMethod refactoring.
2804 Before the analyzer is able to work with the text selections of a method, it
2805 needs to generate them. To do this, it parses the method to obtain a
2806 \type{MethodDeclaration} for it \see{astEclipse}. Then there is a statement
2807 lists creator that creates statements lists of the different groups of
2808 statements in the body of the method declaration. A text selections generator
2809 generates text selections of all the statement lists for the analyzer to work
2812 \paragraph{The statement lists creator}
2813 is responsible for generating lists of statements for all the possible nesting
2814 levels of statements in the method. The statement lists creator is implemented
2815 as an AST visitor \see{astVisitor}. It generates lists of statements by visiting
2816 all the blocks in the method declaration and stores their statements in a
2817 collection of statement lists. In addition, it visits all of the other
2818 statements that can have a statement as a child, such as the different control
2819 structures and the labeled statement.
2821 The switch statement is the only kind of statement that is not straight forward
2822 to obtain the child statements from. It stores all of its children in a flat
2823 list. Its switch case statements are included in this list. This means that
2824 there are potential statement lists between all of these case statements. The
2825 list of statements from a switch statement is therefore traversed, and the
2826 statements between the case statements are grouped as separate lists.
2828 The highlighted part of \myref{lst:grandExample} shows an example of how the
2829 statement lists creator would separate a method body into lists of statements.
2831 \paragraph{The text selections generator} generates text selections for each
2832 list of statements from the statement lists creator. The generator generates a
2833 text selection for every sub-sequence of statements in a list. For a sequence of
2834 statements, the first statement and the last statement span out a text
2837 In practice, the text selections are calculated by only one traversal of the
2838 statement list. There is a set of generated text selections. For each statement,
2839 there is created a temporary set of selections, in addition to a text selection
2840 based on the offset and length of the statement. This text selection is added to
2841 the temporary set. Then the new selection is added with every selection from the
2842 set of generated text selections. These new selections are added to the
2843 temporary set. Then the temporary set of selections is added to the set of
2844 generated text selections. The result of adding two text selections is a new
2845 text selection spanned out by the two addends.
2849 \def\charwidth{5.7pt}
2850 \def\indent{4*\charwidth}
2851 \def\lineheight{\baselineskip}
2852 \def\mintedtop{\lineheight}
2854 \begin{tikzpicture}[overlay, yscale=-1]
2855 \tikzstyle{overlaybox}=[fill=lightgray,opacity=0.2]
2857 \draw[overlaybox] (2*\charwidth,\mintedtop) rectangle
2858 +(18*\charwidth,\lineheight);
2860 \draw[overlaybox] (2*\charwidth,\mintedtop+\lineheight) rectangle
2861 +(18*\charwidth,\lineheight);
2863 \draw[overlaybox] (2*\charwidth,\mintedtop+3*\lineheight) rectangle
2864 +(18*\charwidth,\lineheight);
2866 \draw[overlaybox] (\indent-3*\charwidth,\mintedtop) rectangle
2867 +(20*\charwidth,2*\lineheight);
2869 \draw[overlaybox] (3*\charwidth,\mintedtop+\lineheight) rectangle
2870 +(16*\charwidth,3*\lineheight);
2872 \draw[overlaybox] (\indent,\mintedtop) rectangle
2873 +(14*\charwidth,4*\lineheight);
2875 \begin{minted}{java}
2881 \caption{Example of how the text selections generator would generate text
2882 selections based on a lists of statements. Each highlighted rectangle
2883 represents a text selection.}
2884 \label{lst:textSelectionsExample}
2886 \todoin{fix \myref{lst:textSelectionsExample}? Text only? All
2887 sub-sequences\ldots}
2890 \paragraph{Finding the candidate} for the refactoring is done by analyzing all
2891 the generated text selection with the \type{ExtractAndMoveMethodAnalyzer}
2892 \see{extractAndMoveMethodAnalyzer}. If the analyzer generates a useful result,
2893 an \type{ExtractAndMoveMethodCandidate} is created from it, which is kept in a
2894 list of potential candidates. If no candidates are found, the
2895 \type{NoTargetFoundException} is thrown.
2897 Since only one of the candidates can be chosen, the analyzer must sort out which
2898 candidate to choose. The sorting is done by the static \method{sort} method of
2899 \type{Collections}. The comparison in this sorting is done by an
2900 \type{ExtractAndMoveMethodCandidateComparator}.
2901 \todoin{Write about the
2902 ExtractAndMoveMethodCandidateComparator/FavorNoUnfixesCandidateComparator}
2905 \subsection{The Prefix class}
2906 This class exists mainly for holding data about a prefix, such as the expression
2907 that the prefix represents and the occurrence count of the prefix within a
2908 selection. In addition to this, it has some functionality such as calculating
2909 its sub-prefixes and intersecting it with another prefix. The definition of the
2910 intersection between two prefixes is a prefix representing the longest common
2911 expression between the two.
2913 \subsection{The PrefixSet class}
2914 A prefix set holds elements of type \type{Prefix}. It is implemented with the
2915 help of a \typewithref{java.util}{HashMap} and contains some typical set
2916 operations, but it does not implement the \typewithref{java.util}{Set}
2917 interface, since the prefix set does not need all of the functionality a
2918 \type{Set} requires to be implemented. In addition It needs some other
2919 functionality not found in the \type{Set} interface. So due to the relatively
2920 limited use of prefix sets, and that it almost always needs to be referenced as
2921 such, and not a \type{Set<Prefix>}, it remains as an ad hoc solution to a
2924 There are two ways adding prefixes to a \type{PrefixSet}. The first is through
2925 its \method{add} method. This works like one would expect from a set. It adds
2926 the prefix to the set if it does not already contain the prefix. The other way
2927 is to \emph{register} the prefix with the set. When registering a prefix, if the
2928 set does not contain the prefix, it is just added. If the set contains the
2929 prefix, its count gets incremented. This is how the occurrence count is handled.
2931 The prefix set also computes the set of prefixes that is not enclosing any
2932 prefixes of another set. This is kind of a set difference operation only for
2935 \subsection{Hacking the refactoring undo
2936 history}\label{hacking_undo_history}
2937 \todoin{Where to put this section?}
2939 As an attempt to make multiple subsequent changes to the workspace appear as a
2940 single action (i.e. make the undo changes appear as such), I tried to alter
2941 the undo changes\typeref{org.eclipse.ltk.core.refactoring.Change} in the history
2942 of the refactorings.
2944 My first impulse was to remove the, in this case, last two undo changes from the
2945 undo manager\typeref{org.eclipse.ltk.core.refactoring.IUndoManager} for the
2946 \name{Eclipse} refactorings, and then add them to a composite
2947 change\typeref{org.eclipse.ltk.core.refactoring.CompositeChange} that could be
2948 added back to the manager. The interface of the undo manager does not offer a
2949 way to remove/pop the last added undo change, so a possible solution could be to
2950 decorate\citing{designPatterns} the undo manager, to intercept and collect the
2951 undo changes before delegating to the \method{addUndo}
2952 method\methodref{org.eclipse.ltk.core.refactoring.IUndoManager}{addUndo} of the
2953 manager. Instead of giving it the intended undo change, a null change could be
2954 given to prevent it from making any changes if run. Then one could let the
2955 collected undo changes form a composite change to be added to the manager.
2957 There is a technical challenge with this approach, and it relates to the undo
2958 manager, and the concrete implementation
2959 \typewithref{org.eclipse.ltk.internal.core.refactoring}{UndoManager2}. This
2960 implementation is designed in a way that it is not possible to just add an undo
2961 change, you have to do it in the context of an active
2962 operation\typeref{org.eclipse.core.commands.operations.TriggeredOperations}.
2963 One could imagine that it might be possible to trick the undo manager into
2964 believing that you are doing a real change, by executing a refactoring that is
2965 returning a kind of null change that is returning our composite change of undo
2966 refactorings when it is performed. But this is not the way to go.
2968 Apart from the technical problems with this solution, there is a functional
2969 problem: If it all had worked out as planned, this would leave the undo history
2970 in a dirty state, with multiple empty undo operations corresponding to each of
2971 the sequentially executed refactoring operations, followed by a composite undo
2972 change corresponding to an empty change of the workspace for rounding of our
2973 composite refactoring. The solution to this particular problem could be to
2974 intercept the registration of the intermediate changes in the undo manager, and
2975 only register the last empty change.
2977 Unfortunately, not everything works as desired with this solution. The grouping
2978 of the undo changes into the composite change does not make the undo operation
2979 appear as an atomic operation. The undo operation is still split up into
2980 separate undo actions, corresponding to the changes done by their originating
2981 refactorings. And in addition, the undo actions have to be performed separate in
2982 all the editors involved. This makes it no solution at all, but a step toward
2985 There might be a solution to this problem, but it remains to be found. The
2986 design of the refactoring undo management is partly to be blamed for this, as
2987 it is too complex to be easily manipulated.
2990 \chapter{Analyzing source code in Eclipse}
2992 \section{The Java model}\label{javaModel}
2993 The Java model of \name{Eclipse} is its internal representation of a Java project. It
2994 is light-weight, and has only limited possibilities for manipulating source
2995 code. It is typically used as a basis for the Package Explorer in \name{Eclipse}.
2997 The elements of the Java model are only handles to the underlying elements. This
2998 means that the underlying element of a handle does not need to actually exist.
2999 Hence the user of a handle must always check that it exist by calling the
3000 \method{exists} method of the handle.
3002 The handles with descriptions are listed in \myref{tab:javaModel}, while the
3003 hierarchy of the Java Model is shown in \myref{fig:javaModel}.
3006 \caption{The elements of the Java Model\citing{vogelEclipseJDT2012}.}
3007 \label{tab:javaModel}
3009 % sum must equal number of columns (3)
3010 \begin{tabularx}{\textwidth}{@{} L{0.7} L{1.1} L{1.2} @{}}
3012 \textbf{Project Element} & \textbf{Java Model element} &
3013 \textbf{Description} \\
3015 Java project & \type{IJavaProject} & The Java project which contains all other objects. \\
3017 Source folder /\linebreak[2] binary folder /\linebreak[3] external library &
3018 \type{IPackageFragmentRoot} & Hold source or binary files, can be a folder
3019 or a library (zip / jar file). \\
3021 Each package & \type{IPackageFragment} & Each package is below the
3022 \type{IPackageFragmentRoot}, sub-packages are not leaves of the package,
3023 they are listed directed under \type{IPackageFragmentRoot}. \\
3025 Java Source file & \type{ICompilationUnit} & The Source file is always below
3026 the package node. \\
3028 Types / Fields /\linebreak[3] Methods & \type{IType} / \type{IField}
3029 /\linebreak[3] \type{IMethod} & Types, fields and methods. \\
3037 \begin{tikzpicture}[%
3038 grow via three points={one child at (0,-0.7) and
3039 two children at (0,-0.7) and (0,-1.4)},
3040 edge from parent path={(\tikzparentnode.south west)+(0.5,0) |-
3041 (\tikzchildnode.west)}]
3042 \tikzstyle{every node}=[draw=black,thick,anchor=west]
3043 \tikzstyle{selected}=[draw=red,fill=red!30]
3044 \tikzstyle{optional}=[dashed,fill=gray!50]
3045 \node {\type{IJavaProject}}
3046 child { node {\type{IPackageFragmentRoot}}
3047 child { node {\type{IPackageFragment}}
3048 child { node {\type{ICompilationUnit}}
3049 child { node {\type{IType}}
3050 child { node {\type{\{ IType \}*}}
3051 child { node {\type{\ldots}}}
3054 child { node {\type{\{ IField \}*}}}
3055 child { node {\type{IMethod}}
3056 child { node {\type{\{ IType \}*}}
3057 child { node {\type{\ldots}}}
3062 child { node {\type{\{ IMethod \}*}}}
3071 child { node {\type{\{ IType \}*}}}
3082 child { node {\type{\{ ICompilationUnit \}*}}}
3095 child { node {\type{\{ IPackageFragment \}*}}}
3110 child { node {\type{\{ IPackageFragmentRoot \}*}}}
3113 \caption{The Java model of \name{Eclipse}. ``\type{\{ SomeElement \}*}'' means
3114 ``\type{SomeElement} zero or more times``. For recursive structures,
3115 ``\type{\ldots}'' is used.}
3116 \label{fig:javaModel}
3119 \section{The abstract syntax tree}
3120 \name{Eclipse} is following the common paradigm of using an abstract syntax tree for
3121 source code analysis and manipulation.
3123 When parsing program source code into something that can be used as a foundation
3124 for analysis, the start of the process follows the same steps as in a compiler.
3125 This is all natural, because the way a compiler analyzes code is no different
3126 from how source manipulation programs would do it, except for some properties of
3127 code that is analyzed in the parser, and that they may be differing in what
3128 kinds of properties they analyze. Thus the process of translation source code
3129 into a structure that is suitable for analyzing, can be seen as a kind of
3130 interrupted compilation process \see{fig:interruptedCompilationProcess}.
3135 base/.style={anchor=north, align=center, rectangle, minimum height=1.4cm},
3136 basewithshadow/.style={base, drop shadow, fill=white},
3137 outlined/.style={basewithshadow, draw, rounded corners, minimum
3139 primary/.style={outlined, font=\bfseries},
3140 dashedbox/.style={outlined, dashed},
3141 arrowpath/.style={black, align=center, font=\small},
3142 processarrow/.style={arrowpath, ->, >=angle 90, shorten >=1pt},
3144 \begin{tikzpicture}[node distance=1.3cm and 3cm, scale=1, every
3145 node/.style={transform shape}]
3146 \node[base](AuxNode1){\small source code};
3147 \node[primary, right=of AuxNode1, xshift=-2.5cm](Scanner){Scanner};
3148 \node[primary, right=of Scanner, xshift=0.5cm](Parser){Parser};
3149 \node[dashedbox, below=of Parser](SemanticAnalyzer){Semantic\\Analyzer};
3150 \node[dashedbox, left=of SemanticAnalyzer](SourceCodeOptimizer){Source
3152 \node[dashedbox, below=of SourceCodeOptimizer
3153 ](CodeGenerator){Code\\Generator};
3154 \node[dashedbox, right=of CodeGenerator](TargetCodeOptimizer){Target
3156 \node[base, right=of TargetCodeOptimizer](AuxNode2){};
3158 \draw[processarrow](AuxNode1) -- (Scanner);
3160 \path[arrowpath] (Scanner) -- node [sloped](tokens){tokens}(Parser);
3161 \draw[processarrow](Scanner) -- (tokens) -- (Parser);
3163 \path[arrowpath] (Parser) -- node (syntax){syntax
3164 tree}(SemanticAnalyzer);
3165 \draw[processarrow](Parser) -- (syntax) -- (SemanticAnalyzer);
3167 \path[arrowpath] (SemanticAnalyzer) -- node
3168 [sloped](annotated){annotated\\tree}(SourceCodeOptimizer);
3169 \draw[processarrow, dashed](SemanticAnalyzer) -- (annotated) --
3170 (SourceCodeOptimizer);
3172 \path[arrowpath] (SourceCodeOptimizer) -- node
3173 (intermediate){intermediate code}(CodeGenerator);
3174 \draw[processarrow, dashed](SourceCodeOptimizer) -- (intermediate) --
3177 \path[arrowpath] (CodeGenerator) -- node [sloped](target1){target
3178 code}(TargetCodeOptimizer);
3179 \draw[processarrow, dashed](CodeGenerator) -- (target1) --
3180 (TargetCodeOptimizer);
3182 \path[arrowpath](TargetCodeOptimizer) -- node [sloped](target2){target
3184 \draw[processarrow, dashed](TargetCodeOptimizer) -- (target2) (AuxNode2);
3186 \caption{Interrupted compilation process. {\footnotesize (Full compilation
3187 process borrowed from \emph{Compiler construction: principles and practice}
3188 by Kenneth C. Louden\citing{louden1997}.)}}
3189 \label{fig:interruptedCompilationProcess}
3192 The process starts with a \emph{scanner}, or lexer. The job of the scanner is to
3193 read the source code and divide it into tokens for the parser. Therefore, it is
3194 also sometimes called a tokenizer. A token is a logical unit, defined in the
3195 language specification, consisting of one or more consecutive characters. In
3196 the Java language the tokens can for instance be the \var{this} keyword, a curly
3197 bracket \var{\{} or a \var{nameToken}. It is recognized by the scanner on the
3198 basis of something equivalent of a regular expression. This part of the process
3199 is often implemented with the use of a finite automata. In fact, it is common to
3200 specify the tokens in regular expressions, which in turn are translated into a
3201 finite automata lexer. This process can be automated.
3203 The program component used to translate a stream of tokens into something
3204 meaningful, is called a parser. A parser is fed tokens from the scanner and
3205 performs an analysis of the structure of a program. It verifies that the syntax
3206 is correct according to the grammar rules of a language, that are usually
3207 specified in a context-free grammar, and often in a variant of the
3208 \name{Backus--Naur Form}. The result coming from the parser is in the form of an
3209 \emph{Abstract Syntax Tree}, AST for short. It is called \emph{abstract},
3210 because the structure does not contain all of the tokens produced by the
3211 scanner. It only contains logical constructs, and because it forms a tree, all
3212 kinds of parentheses and brackets are implicit in the structure. It is this AST
3213 that is used when performing the semantic analysis of the code.
3215 As an example we can think of the expression \code{(5 + 7) * 2}. The root of
3216 this tree would in \name{Eclipse} be an \type{InfixExpression} with the operator
3217 \var{TIMES}, and a left operand, which is also an \type{InfixExpression} with
3218 the operator \var{PLUS}. The left operand \type{InfixExpression}, has in turn a
3219 left operand of type \type{NumberLiteral} with the value \var{``5''} and a right
3220 operand \type{NumberLiteral} with the value \var{``7''}. The root will have a
3221 right operand of type \type{NumberLiteral} and value \var{``2''}. The AST for
3222 this expression is illustrated in \myref{fig:astInfixExpression}.
3224 Contrary to the Java Model, an abstract syntax tree is a heavy-weight
3225 representation of source code. It contains information about properties like
3226 type bindings for variables and variable bindings for names.
3231 \begin{tikzpicture}[scale=0.8]
3232 \tikzset{level distance=40pt}
3233 \tikzset{sibling distance=5pt}
3234 \tikzstyle{thescale}=[scale=0.8]
3235 \tikzset{every tree node/.style={align=center}}
3236 \tikzset{edge from parent/.append style={thick}}
3237 \tikzstyle{inode}=[rectangle,rounded corners,draw,fill=lightgray,drop
3238 shadow,align=center]
3239 \tikzset{every internal node/.style={inode}}
3240 \tikzset{every leaf node/.style={draw=none,fill=none}}
3242 \Tree [.\type{InfixExpression} [.\type{InfixExpression}
3243 [.\type{NumberLiteral} \var{``5''} ] [.\type{Operator} \var{PLUS} ]
3244 [.\type{NumberLiteral} \var{``7''} ] ]
3245 [.\type{Operator} \var{TIMES} ]
3246 [.\type{NumberLiteral} \var{``2''} ]
3249 \caption{The abstract syntax tree for the expression \code{(5 + 7) * 2}.}
3250 \label{fig:astInfixExpression}
3253 \subsection{The AST in Eclipse}\label{astEclipse}
3254 In \name{Eclipse}, every node in the AST is a child of the abstract superclass
3255 \typewithref{org.eclipse.jdt.core.dom}{ASTNode}. Every \type{ASTNode}, among a
3256 lot of other things, provides information about its position and length in the
3257 source code, as well as a reference to its parent and to the root of the tree.
3259 The root of the AST is always of type \type{CompilationUnit}. It is not the same
3260 as an instance of an \type{ICompilationUnit}, which is the compilation unit
3261 handle of the Java model. The children of a \type{CompilationUnit} is an
3262 optional \type{PackageDeclaration}, zero or more nodes of type
3263 \type{ImportDecaration} and all its top-level type declarations that has node
3264 types \type{AbstractTypeDeclaration}.
3266 An \type{AbstractType\-Declaration} can be one of the types
3267 \type{AnnotationType\-Declaration}, \type{Enum\-Declaration} or
3268 \type{Type\-Declaration}. The children of an \type{AbstractType\-Declaration}
3269 must be a subtype of a \type{BodyDeclaration}. These subtypes are:
3270 \type{AnnotationTypeMember\-Declaration}, \type{EnumConstant\-Declaration},
3271 \type{Field\-Declaration}, \type{Initializer} and \type{Method\-Declaration}.
3273 Of the body declarations, the \type{Method\-Declaration} is the most interesting
3274 one. Its children include lists of modifiers, type parameters, parameters and
3275 exceptions. It has a return type node and a body node. The body, if present, is
3276 of type \type{Block}. A \type{Block} is itself a \type{Statement}, and its
3277 children is a list of \type{Statement} nodes.
3279 There are too many types of the abstract type \type{Statement} to list up, but
3280 there exists a subtype of \type{Statement} for every statement type of Java, as
3281 one would expect. This also applies to the abstract type \type{Expression}.
3282 However, the expression \type{Name} is a little special, since it is both used
3283 as an operand in compound expressions, as well as for names in type declarations
3286 There is an overview of some of the structure of an \name{Eclipse} AST in
3287 \myref{fig:astEclipse}.
3291 \begin{tikzpicture}[scale=0.8]
3292 \tikzset{level distance=50pt}
3293 \tikzset{sibling distance=5pt}
3294 \tikzstyle{thescale}=[scale=0.8]
3295 \tikzset{every tree node/.style={align=center}}
3296 \tikzset{edge from parent/.append style={thick}}
3297 \tikzstyle{inode}=[rectangle,rounded corners,draw,fill=lightgray,drop
3298 shadow,align=center]
3299 \tikzset{every internal node/.style={inode}}
3300 \tikzset{every leaf node/.style={draw=none,fill=none}}
3302 \Tree [.\type{CompilationUnit} [.\type{[ PackageDeclaration ]} [.\type{Name} ]
3303 [.\type{\{ Annotation \}*} ] ]
3304 [.\type{\{ ImportDeclaration \}*} [.\type{Name} ] ]
3305 [.\type{\{ AbstractTypeDeclaration \}+} [.\node(site){\type{\{
3306 BodyDeclaration \}*}}; ] [.\type{SimpleName} ] ]
3308 \begin{scope}[shift={(0.5,-6)}]
3309 \node[inode,thescale](root){\type{MethodDeclaration}};
3310 \node[inode,thescale](modifiers) at (4.5,-5){\type{\{ IExtendedModifier \}*}
3311 \\ {\footnotesize (Of type \type{Modifier} or \type{Annotation})}};
3312 \node[inode,thescale](typeParameters) at (-6,-3.5){\type{\{ TypeParameter
3314 \node[inode,thescale](parameters) at (-5,-5){\type{\{
3315 SingleVariableDeclaration \}*} \\ {\footnotesize (Parameters)}};
3316 \node[inode,thescale](exceptions) at (5,-3){\type{\{ Name \}*} \\
3317 {\footnotesize (Exceptions)}};
3318 \node[inode,thescale](return) at (-6.5,-2){\type{Type} \\ {\footnotesize
3320 \begin{scope}[shift={(0,-5)}]
3321 \Tree [.\node(body){\type{[ Block ]} \\ {\footnotesize (Body)}};
3322 [.\type{\{ Statement \}*} [.\type{\{ Expression \}*} ]
3323 [.\type{\{ Statement \}*} [.\type{\ldots} ]]
3328 \draw[->,>=triangle 90,shorten >=1pt](root.east)..controls +(east:2) and
3329 +(south:1)..(site.south);
3331 \draw (root.south) -- (modifiers);
3332 \draw (root.south) -- (typeParameters);
3333 \draw (root.south) -- ($ (parameters.north) + (2,0) $);
3334 \draw (root.south) -- (exceptions);
3335 \draw (root.south) -- (return);
3336 \draw (root.south) -- (body);
3339 \caption{The format of the abstract syntax tree in \name{Eclipse}.}
3340 \label{fig:astEclipse}
3343 \section{The ASTVisitor}\label{astVisitor}
3344 So far, the only thing that has been addressed is how the data that is going to
3345 be the basis for our analysis is structured. Another aspect of it is how we are
3346 going to traverse the AST to gather the information we need, so we can conclude
3347 about the properties we are analyzing. It is of course possible to start at the
3348 top of the tree, and manually search through its nodes for the ones we are
3349 looking for, but that is a bit inconvenient. To be able to efficiently utilize
3350 such an approach, we would need to make our own framework for traversing the
3351 tree and visiting only the types of nodes we are after. Luckily, this
3352 functionality is already provided in \name{Eclipse}, by its
3353 \typewithref{org.eclipse.jdt.core.dom}{ASTVisitor}.
3355 The \name{Eclipse} AST, together with its \type{ASTVisitor}, follows the
3356 \pattern{Visitor} pattern\citing{designPatterns}. The intent of this design
3357 pattern is to facilitate extending the functionality of classes without touching
3358 the classes themselves.
3360 Let us say that there is a class hierarchy of elements. These elements all have
3361 a method \method{accept(Visitor visitor)}. In its simplest form, the
3362 \method{accept} method just calls the \method{visit} method of the visitor with
3363 itself as an argument, like this: \code{visitor.visit(this)}. For the visitors
3364 to be able to extend the functionality of all the classes in the elements
3365 hierarchy, each \type{Visitor} must have one visit method for each concrete
3366 class in the hierarchy. Say the hierarchy consists of the concrete classes
3367 \type{ConcreteElementA} and \type{ConcreteElementB}. Then each visitor must have
3368 the (possibly empty) methods \method{visit(ConcreteElementA element)} and
3369 \method{visit(ConcreteElementB element)}. This scenario is depicted in
3370 \myref{fig:visitorPattern}.
3374 \tikzstyle{abstract}=[rectangle, draw=black, fill=white, drop shadow, text
3375 centered, anchor=north, text=black, text width=6cm, every one node
3376 part/.style={align=center, font=\bfseries\itshape}]
3377 \tikzstyle{concrete}=[rectangle, draw=black, fill=white, drop shadow, text
3378 centered, anchor=north, text=black, text width=6cm]
3379 \tikzstyle{inheritarrow}=[->, >=open triangle 90, thick]
3380 \tikzstyle{commentarrow}=[->, >=angle 90, dashed]
3381 \tikzstyle{line}=[-, thick]
3382 \tikzset{every one node part/.style={align=center, font=\bfseries}}
3383 \tikzset{every second node part/.style={align=center, font=\ttfamily}}
3385 \begin{tikzpicture}[node distance=1cm, scale=0.8, every node/.style={transform
3387 \node (Element) [abstract, rectangle split, rectangle split parts=2]
3389 \nodepart{one}{Element}
3390 \nodepart{second}{+accept(visitor: Visitor)}
3392 \node (AuxNode01) [text width=0, minimum height=2cm, below=of Element] {};
3393 \node (ConcreteElementA) [concrete, rectangle split, rectangle split
3394 parts=2, left=of AuxNode01]
3396 \nodepart{one}{ConcreteElementA}
3397 \nodepart{second}{+accept(visitor: Visitor)}
3399 \node (ConcreteElementB) [concrete, rectangle split, rectangle split
3400 parts=2, right=of AuxNode01]
3402 \nodepart{one}{ConcreteElementB}
3403 \nodepart{second}{+accept(visitor: Visitor)}
3406 \node[comment, below=of ConcreteElementA] (CommentA) {visitor.visit(this)};
3408 \node[comment, below=of ConcreteElementB] (CommentB) {visitor.visit(this)};
3410 \node (AuxNodeX) [text width=0, minimum height=1cm, below=of AuxNode01] {};
3412 \node (Visitor) [abstract, rectangle split, rectangle split parts=2,
3415 \nodepart{one}{Visitor}
3416 \nodepart{second}{+visit(ConcreteElementA)\\+visit(ConcreteElementB)}
3418 \node (AuxNode02) [text width=0, minimum height=2cm, below=of Visitor] {};
3419 \node (ConcreteVisitor1) [concrete, rectangle split, rectangle split
3420 parts=2, left=of AuxNode02]
3422 \nodepart{one}{ConcreteVisitor1}
3423 \nodepart{second}{+visit(ConcreteElementA)\\+visit(ConcreteElementB)}
3425 \node (ConcreteVisitor2) [concrete, rectangle split, rectangle split
3426 parts=2, right=of AuxNode02]
3428 \nodepart{one}{ConcreteVisitor2}
3429 \nodepart{second}{+visit(ConcreteElementA)\\+visit(ConcreteElementB)}
3433 \draw[inheritarrow] (ConcreteElementA.north) -- ++(0,0.7) -|
3435 \draw[line] (ConcreteElementA.north) -- ++(0,0.7) -|
3436 (ConcreteElementB.north);
3438 \draw[inheritarrow] (ConcreteVisitor1.north) -- ++(0,0.7) -|
3440 \draw[line] (ConcreteVisitor1.north) -- ++(0,0.7) -|
3441 (ConcreteVisitor2.north);
3443 \draw[commentarrow] (CommentA.north) -- (ConcreteElementA.south);
3444 \draw[commentarrow] (CommentB.north) -- (ConcreteElementB.south);
3448 \caption{The Visitor Pattern.}
3449 \label{fig:visitorPattern}
3452 The use of the visitor pattern can be appropriate when the hierarchy of elements
3453 is mostly stable, but the family of operations over its elements is constantly
3454 growing. This is clearly the case for the \name{Eclipse} AST, since the
3455 hierarchy for the type \type{ASTNode} is very stable, but the functionality of
3456 its elements is extended every time someone need to operate on the AST. Another
3457 aspect of the \name{Eclipse} implementation is that it is a public API, and the
3458 visitor pattern is an easy way to provide access to the nodes in the tree.
3460 The version of the visitor pattern implemented for the AST nodes in \name{Eclipse} also
3461 provides an elegant way to traverse the tree. It does so by following the
3462 convention that every node in the tree first let the visitor visit itself,
3463 before it also makes all its children accept the visitor. The children are only
3464 visited if the visit method of their parent returns \var{true}. This pattern
3465 then makes for a prefix traversal of the AST. If postfix traversal is desired,
3466 the visitors also have \method{endVisit} methods for each node type, which is
3467 called after the \method{visit} method for a node. In addition to these visit
3468 methods, there are also the methods \method{preVisit(ASTNode)},
3469 \method{postVisit(ASTNode)} and \method{preVisit2(ASTNode)}. The
3470 \method{preVisit} method is called before the type-specific \method{visit}
3471 method. The \method{postVisit} method is called after the type-specific
3472 \method{endVisit}. The type specific \method{visit} is only called if
3473 \method{preVisit2} returns \var{true}. Overriding the \method{preVisit2} is also
3474 altering the behavior of \method{preVisit}, since the default implementation is
3475 responsible for calling it.
3477 An example of a trivial \type{ASTVisitor} is shown in
3478 \myref{lst:astVisitorExample}.
3481 \begin{minted}{java}
3482 public class CollectNamesVisitor extends ASTVisitor {
3483 Collection<Name> names = new LinkedList<Name>();
3486 public boolean visit(QualifiedName node) {
3492 public boolean visit(SimpleName node) {
3498 \caption{An \type{ASTVisitor} that visits all the names in a subtree and adds
3499 them to a collection, except those names that are children of any
3500 \type{QualifiedName}.}
3501 \label{lst:astVisitorExample}
3504 \section{Property collectors}\label{propertyCollectors}
3505 The prefixes and unfixes are found by property
3506 collectors\typeref{no.uio.ifi.refaktor.extractors.collectors.PropertyCollector}.
3507 A property collector is of the \type{ASTVisitor} type, and thus visits nodes of
3508 type \type{ASTNode} of the abstract syntax tree \see{astVisitor}.
3510 \subsection{The PrefixesCollector}
3511 The \typewithref{no.uio.ifi.refaktor.extractors.collectors}{PrefixesCollector}
3512 finds prefixes that makes up the basis for calculating move targets for the
3513 \refa{Extract and Move Method} refactoring. It visits expression
3514 statements\typeref{org.eclipse.jdt.core.dom.ExpressionStatement} and creates
3515 prefixes from its expressions in the case of method invocations. The prefixes
3516 found are registered with a prefix set, together with all its sub-prefixes.
3518 \subsection{The UnfixesCollector}\label{unfixes}
3519 The \typewithref{no.uio.ifi.refaktor.extractors.collectors}{UnfixesCollector}
3520 finds unfixes within a selection.
3521 \todoin{Give more technical detail?}
3523 \section{Checkers}\label{checkers}
3524 The checkers are a range of classes that checks that text selections comply
3525 with certain criteria. All checkers operates under the assumption that the code
3526 they check is free from compilation errors. If a
3527 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{Checker} fails, it throws a
3528 \type{CheckerException}. The checkers are managed by the
3529 \type{LegalStatementsChecker}, which does not, in fact, implement the
3530 \type{Checker} interface. It does, however, run all the checkers registered with
3531 it, and reports that all statements are considered legal if no
3532 \type{CheckerException} is thrown. Many of the checkers either extends the
3533 \type{PropertyCollector} or utilizes one or more property collectors to verify
3534 some criteria. The checkers registered with the \type{LegalStatementsChecker}
3535 are described next. They are run in the order presented below.
3537 \subsection{The CallToProtectedOrPackagePrivateMethodChecker}
3538 This checker is used to check that at selection does not contain a call to a
3539 method that is protected or package-private. Such a method either has the access
3540 modifier \code{protected} or it has no access modifier.
3542 The workings of the \type{CallToProtectedOrPackagePrivateMethod\-Checker} is
3543 very simple. It looks for calls to methods that are either protected or
3544 package-private within the selection, and throws an
3545 \type{IllegalExpressionFoundException} if one is found.
3547 \subsection{The DoubleClassInstanceCreationChecker}
3548 The \type{DoubleClassInstanceCreationChecker} checks that there are no double
3549 class instance creations where the inner constructor call takes an argument that
3550 is built up using field references.
3552 The checker visits all nodes of type \type{ClassInstanceCreation} within a
3553 selection. For all of these nodes, if its parent also is a class instance
3554 creation, it accepts a visitor that throws a
3555 \type{IllegalExpressionFoundException} if it encounters a name that is a field
3558 \subsection{The InstantiationOfNonStaticInnerClassChecker}
3559 The \type{InstantiationOfNonStaticInnerClassChecker} checks that selections
3560 do not contain instantiations of non-static inner classes. The
3561 \type{MoveInstanceMethodProcessor} in \name{Eclipse} does not handle such
3562 instantiations gracefully when moving a method. This problem is also related to
3563 bug\ldots \todoin{File Eclipse bug report}
3565 \subsection{The EnclosingInstanceReferenceChecker}
3566 The purpose of this checker is to verify that the names in a text selection are
3567 not referencing any enclosing instances. In theory, the underlying problem could
3568 be solved in some situations, but our dependency on the
3569 \type{MoveInstanceMethodProcessor} prevents this.
3572 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{EnclosingInstanceReferenceChecker}
3573 is a modified version of the
3574 \typewithref{org.eclipse.jdt.internal.corext.refactoring.structure.MoveInstanceMethod\-Processor}{EnclosingInstanceReferenceFinder}
3575 from the \type{MoveInstanceMethodProcessor}. Wherever the
3576 \type{EnclosingInstanceReferenceFinder} would create a fatal error status, the
3577 checker will throw a \type{CheckerException}.
3579 The checker works by first finding all of the enclosing types of a selection.
3580 Thereafter, it visits all the simple names of the selection to check that they
3581 are not references to variables or methods declared in any of the enclosing
3582 types. In addition, the checker visits \var{this}-expressions to verify that no
3583 such expressions are qualified with any name.
3585 \subsection{The ReturnStatementsChecker}\label{returnStatementsChecker}
3586 The checker for return statements is meant to verify that a text selection is
3587 consistent regarding return statements.
3589 If the selection is free from return statements, then the checker validates. So
3590 this is the first thing the checker investigates.
3592 If the checker proceeds any further, it is because the selection contains one
3593 or more return statements. The next test is therefore to check if the last
3594 statement of the selection ends in either a return or a throw statement. The
3595 responsibility for checking that the last statement of the selection eventually
3596 ends in a return or throw statement, is put on the
3597 \type{LastStatementOfSelectionEndsInReturnOrThrowChecker}. For every node
3598 visited, if the node is a statement, it does a test to see if the statement is a
3599 return, a throw or if it is an implicit return statement. If this is the case,
3600 no further checking is done. This checking is done in the \code{preVisit2}
3601 method \see{astVisitor}. If the node is not of a type that is being handled by
3602 its type-specific visit method, the checker performs a simple test. If the node
3603 being visited is not the last statement of its parent that is also enclosed by
3604 the selection, an \type{IllegalStatementFoundException} is thrown. This ensures
3605 that all statements are taken care of, one way or the other. It also ensures
3606 that the checker is conservative in the way it checks for legality of the
3609 To examine if a statement is an implicit return statement, the checker first
3610 finds the last statement declared in its enclosing method. If this statement is
3611 the same as the one under investigation, it is considered an implicit return
3612 statement. If the statements are not the same, the checker does a search to see
3613 if the statement examined is also the last statement of the method that can be
3614 reached. This includes the last statement of a block statement, a labeled
3615 statement, a synchronized statement or a try statement, that in turn is the last
3616 statement enclosed by one of the statement types listed. This search goes
3617 through all the parents of a statement until a statement is found that is not
3618 one of the mentioned acceptable parent statements. If the search ends in a
3619 method declaration, then the statement is considered to be the last reachable
3620 statement of the method, and thus it is an implicit return statement.
3622 There are two kinds of statements that are handled explicitly: If-statements and
3623 try-statements. Block, labeled and do-statements are handled by fall-through to
3626 If-statements are handled explicitly by overriding their type-specific visit
3627 method. If the then-part does not contain any return or throw statements an
3628 \type{IllegalStatementFoundException} is thrown. If it does contain a return or
3629 throw, its else-part is checked. If the else-part is non-existent, or it does
3630 not contain any return or throw statements an exception is thrown. If no
3631 exception is thrown while visiting the if-statement, its children are visited.
3633 A try-statement is checked very similar to an if-statement. Its body must
3634 contain a return or throw. The same applies to its catch clauses and finally
3635 body. Failure to validate produces an \type{IllegalStatementFoundException}.
3637 If the checker does not complain at any point, the selection is considered valid
3638 with respect to return statements.
3640 \subsection{The AmbiguousReturnValueChecker}
3641 This checker verifies that there are no ambiguous return values in a selection.
3643 First, the checker needs to collect some data. Those data are the binding keys
3644 for all simple names that are assigned to within the selection, including
3645 variable declarations, but excluding fields. The checker also finds out whether
3646 a return statement is found in the selection or not. No further checks of return
3647 statements are needed, since, at this point, the selection is already checked
3648 for illegal return statements \see{returnStatementsChecker}.
3650 After the binding keys of the assignees are collected, the checker searches the
3651 part of the enclosing method that is after the selection for references whose
3652 binding keys are among the collected keys. If more than one unique referral is
3653 found, or only one referral is found, but the selection also contains a return
3654 statement, we have a situation with an ambiguous return value, and an exception
3657 %\todoin{Explain why we do not need to consider variables assigned inside
3658 %local/anonymous classes. (The referenced variables need to be final and so
3661 \subsection{The IllegalStatementsChecker}
3662 This checker is designed to check for illegal statements.
3664 Notice that labels in break and continue statements need some special treatment.
3665 Since a label does not have any binding information, we have to search upwards
3666 in the AST to find the \type{LabeledStatement} that corresponds to the label
3667 from the break or continue statement, and check that it is contained in the
3668 selection. If the break or continue statement does not have a label attached to
3669 it, it is checked that its innermost enclosing loop or switch statement (break
3670 statements only) also is contained in the selection.
3672 \chapter{Technicalities}
3674 \section{Source code organization}
3675 All the parts of this master's project are under version control with
3676 \name{Git}\footnote{\url{http://git-scm.com/}}.
3678 The software written is organized as some \name{Eclipse} plugins. Writing a plugin is
3679 the natural way to utilize the API of \name{Eclipse}. This also makes it possible to
3680 provide a user interface to manually run operations on selections in program
3681 source code or whole projects/packages.
3683 When writing a plugin in \name{Eclipse}, one has access to resources such as the
3684 current workspace, the open editor and the current selection.
3686 The thesis work is contained in the following Eclipse projects:
3689 \item[no.uio.ifi.refaktor] \hfill \\ This is the main Eclipse plugin
3690 project, and contains all of the business logic for the plugin.
3692 \item[no.uio.ifi.refaktor.tests] \hfill \\
3693 This project contains the tests for the main plugin.
3695 \item[no.uio.ifi.refaktor.examples] \hfill \\
3696 Contains example code used in testing. It also contains code for managing
3697 this example code, such as creating an Eclipse project from it before a test
3700 \item[no.uio.ifi.refaktor.benchmark] \hfill \\
3701 This project contains code for running search based versions of the
3702 composite refactoring over selected Eclipse projects.
3704 \item[no.uio.ifi.refaktor.releng] \hfill \\
3705 Contains the rmap, queries and target definitions needed by Buckminster on
3706 the Jenkins continuous integration server.
3710 \subsection{The no.uio.ifi.refaktor project}
3712 \subsubsection{no.uio.ifi.refaktor.analyze}
3713 This package, and its sub-packages, contains code that is used for analyzing
3714 Java source code. The most important sub-packages are presented below.
3717 \item[no.uio.ifi.refaktor.analyze.analyzers] \hfill \\
3718 This package contains source code analyzers. These are usually responsible
3719 for analyzing text selections or running specialized analyzers for different
3720 kinds of entities. Their structures are often hierarchical. This means that
3721 you have an analyzer for text selections, that in turn is utilized by an
3722 analyzer that analyzes all the selections of a method. Then there are
3723 analyzers for analyzing all the methods of a type, all the types of a
3724 compilation unit, all the compilation units of a package, and, at last, all
3725 of the packages in a project.
3727 \item[no.uio.ifi.refaktor.analyze.checkers] \hfill \\
3728 A package containing checkers. The checkers are classes used to validate
3729 that a selection can be further analyzed and chosen as a candidate for a
3730 refactoring. Invalidating properties can be such as usage of inner classes
3731 or the need for multiple return values.
3733 \item[no.uio.ifi.refaktor.analyze.collectors] \hfill \\
3734 This package contains the property collectors. Collectors are used to gather
3735 properties from a text selection. This is mostly properties regarding
3736 referenced names and their occurrences. It is these properties that make up
3737 the basis for finding the best candidates for a refactoring.
3740 \subsubsection{no.uio.ifi.refaktor.change}
3741 This package, and its sub-packages, contains functionality for manipulate source
3745 \item[no.uio.ifi.refaktor.change.changers] \hfill \\
3746 This package contains source code changers. They are used to glue together
3747 the analysis of source code and the actual execution of the changes.
3749 \item[no.uio.ifi.refaktor.change.executors] \hfill \\
3750 The executors that are responsible for making concrete changes are found in
3751 this package. They are mostly used to create and execute one or more Eclipse
3754 \item[no.uio.ifi.refaktor.change.processors] \hfill \\
3755 Contains a refactoring processor for the \MoveMethod refactoring. The code
3756 is stolen and modified to fix a bug. The related bug is described in
3757 \myref{eclipse_bug_429416}.
3761 \subsubsection{no.uio.ifi.refaktor.handlers}
3762 This package contains handlers for the commands defined in the plugin manifest.
3764 \subsubsection{no.uio.ifi.refaktor.prefix}
3765 This package contains the \type{Prefix} type that is the data representation of
3766 the prefixes found by the \type{PrefixesCollector}. It also contains the prefix
3767 set for storing and working with prefixes.
3769 \subsubsection{no.uio.ifi.refaktor.statistics}
3770 The package contains statistics functionality. Its heart is the statistics
3771 aspect that is responsible for gathering statistics during the execution of the
3772 \ExtractAndMoveMethod refactoring.
3775 \item[no.uio.ifi.refaktor.statistics.reports] \hfill \\
3776 This package contains a simple framework for generating reports from the
3777 statistics data generated by the aspect. Currently, the only available
3778 report type is a simple text report.
3783 \subsubsection{no.uio.ifi.refaktor.textselection}
3784 This package contains the two custom text selections that are used extensively
3785 throughout the project. One of them is just a subclass of the other, to support
3786 the use of the memento pattern to optimize the memory usage during benchmarking.
3788 \subsubsection{no.uio.ifi.refaktor.debugging}
3789 The package contains a debug utility class. I addition to this, the package
3790 \code{no.uio.ifi.refaktor.utils.aspects} contains a couple of aspects used for
3793 \subsubsection{no.uio.ifi.refaktor.utils}
3794 Utility package that contains all the functionality that has to do with parsing
3795 of source code. It also has utility classes for looking up handles to methods
3796 and types et cetera.
3799 \item[no.uio.ifi.refaktor.utils.caching] \hfill \\
3800 This package contains the caching manager for compilation units, along with
3801 classes for different caching strategies.
3803 \item[no.uio.ifi.refaktor.utils.nullobjects] \hfill \\
3804 Contains classes for creating different null objects. Most of the classes
3805 are used to represent null objects of different handle types. These null
3806 objects are returned from various utility classes instead of returning a
3807 \var{null} value when other values are not available.
3811 \section{Continuous integration}
3812 The continuous integration server
3813 \name{Jenkins}\footnote{\url{http://jenkins-ci.org/}} has been set up for the
3814 project\footnote{A work mostly done by the supervisor.}. It is used as a way to
3815 run tests and perform code coverage analysis.
3817 To be able to build the \name{Eclipse} plugins and run tests for them with Jenkins, the
3818 component assembly project
3819 \name{Buckminster}\footnote{\url{http://www.eclipse.org/buckminster/}} is used,
3820 through its plugin for Jenkins. Buckminster provides for a way to specify the
3821 resources needed for building a project and where and how to find them.
3822 Buckminster also handles the setup of a target environment to run the tests in.
3823 All this is needed because the code to build depends on an \name{Eclipse}
3824 installation with various plugins.
3826 \subsection{Problems with AspectJ}
3827 The Buckminster build worked fine until introducing AspectJ into the project.
3828 When building projects using AspectJ, there are some additional steps that need
3829 to be performed. First of all, the aspects themselves must be compiled. Then the
3830 aspects need to be woven with the classes they affect. This demands a process
3831 that does multiple passes over the source code.
3833 When using AspectJ with \name{Eclipse}, the specialized compilation and the
3834 weaving can be handled by the \name{AspectJ Development
3835 Tools}\footnote{\url{https://www.eclipse.org/ajdt/}}. This works all fine, but
3836 it complicates things when trying to build a project depending on \name{Eclipse}
3837 plugins outside of \name{Eclipse}. There is supposed to be a way to specify a
3838 compiler adapter for javac, together with the file extensions for the file types
3839 it shall operate. The AspectJ compiler adapter is called
3840 \typewithref{org.aspectj.tools.ant.taskdefs}{Ajc11CompilerAdapter}, and it works
3841 with files that has the extensions \code{*.java} and \code{*.aj}. I tried to
3842 setup this in the build properties file for the project containing the aspects,
3843 but to no avail. The project containing the aspects does not seem to be built at
3844 all, and the projects that depend on it complain that they cannot find certain
3847 I then managed to write an \name{Ant}\footnote{\url{https://ant.apache.org/}}
3848 build file that utilizes the AspectJ compiler adapter, for the
3849 \code{no.uio.ifi.refaktor} plugin. The problem was then that it could no longer
3850 take advantage of the environment set up by Buckminster. The solution to this
3851 particular problem was of a ``hacky'' nature. It involves exporting the plugin
3852 dependencies for the project to an Ant build file, and copy the exported path
3853 into the existing build script. But then the Ant script needs to know where the
3854 local \name{Eclipse} installation is located. This is no problem when building
3855 on a local machine, but to utilize the setup done by Buckminster is a problem
3856 still unsolved. To get the classpath for the build setup correctly, and here
3857 comes the most ``hacky'' part of the solution, the Ant script has a target for
3858 copying the classpath elements into a directory relative to the project
3859 directory and checking it into Git. When no \code{ECLIPSE\_HOME} property is set
3860 while running Ant, the script uses the copied plugins instead of the ones
3861 provided by the \name{Eclipse} installation when building the project. This
3862 obviously creates some problems with maintaining the list of dependencies in the
3863 Ant file, as well as remembering to copy the plugins every time the list of
3864 dependencies changes.
3866 The Ant script described above is run by Jenkins before the Buckminster setup
3867 and build. When setup like this, the Buckminster build succeeds for the projects
3868 not using AspectJ, and the tests are run as normal. This is all good, but it
3869 feels a little scary, since the reason for Buckminster not working with AspectJ
3872 The problems with building with AspectJ on the Jenkins server lasted for a
3873 while, before they were solved. This is reflected in the ``Test Result Trend''
3874 and ``Code Coverage Trend'' reported by Jenkins.
3876 \chapter{Benchmarking}\label{sec:benchmarking}
3877 This part of the master's project is located in the \name{Eclipse} project
3878 \code{no.uio.ifi.refaktor.benchmark}. The purpose of it is to run the equivalent
3879 of the \type{SearchBasedExtractAndMoveMethodChanger}
3880 \see{searchBasedExtractAndMoveMethodChanger} over a larger software project,
3881 both to test its robustness but also its effect on different software metrics.
3883 \section{The benchmark setup}
3884 The benchmark itself is set up as a \name{JUnit} test case. This is a convenient
3885 setup, and utilizes the \name{JUnit Plugin Test Launcher}. This provides us with
3886 a fully functional \name{Eclipse} workbench. Most importantly, this gives us
3887 access to the Java Model of \name{Eclipse} \see{javaModel}.
3889 \subsection{The ProjectImporter}
3890 The Java project that is going to be used as the data for the benchmark, must be
3891 imported into the JUnit workspace. This is done by the
3892 \typewithref{no.uio.ifi.refaktor.benchmark}{ProjectImporter}. The importer
3893 requires the absolute path to the project description file. This file is named
3894 \code{.project} and is located at the root of the project directory.
3896 The project description is loaded to find the name of the project to be
3897 imported. The project that shall be the destination for the import is created in
3898 the workspace, on the base of the name from the description. Then an import
3899 operation is created, based on both the source and destination information. The
3900 import operation is run to perform the import.
3902 I have found no simple API call to accomplish what the importer does, which
3903 tells me that it may not be too many people performing this particular action.
3904 The solution to the problem was found on \name{Stack
3905 Overflow}\footnote{\url{https://stackoverflow.com/questions/12401297}}. It
3906 contains enough dirty details to be considered inconvenient to use, if not
3907 wrapping it in a class like my \type{ProjectImporter}. One would probably have
3908 to delve into the source code for the import wizard to find out how the import
3909 operation works, if no one had already done it.
3911 \section{Statistics}
3912 Statistics for the analysis and changes is captured by the
3913 \typewithref{no.uio.ifi.refaktor.aspects}{StatisticsAspect}. This an
3914 \emph{aspect} written in \name{AspectJ}.
3916 \subsection{AspectJ}
3917 \name{AspectJ}\footnote{\url{http://eclipse.org/aspectj/}} is an extension to
3918 the Java language, and facilitates combining aspect-oriented programming with
3919 the object-oriented programming in Java.
3921 Aspect-oriented programming is a programming paradigm that is meant to isolate
3922 so-called \emph{cross-cutting concerns} into their own modules. These
3923 cross-cutting concerns are functionalities that span over multiple classes, but
3924 may not belong naturally in any of them. It can be functionality that does not
3925 concern the business logic of an application, and thus may be a burden when
3926 entangled with parts of the source code it does not really belong. Examples
3927 include logging, debugging, optimization and security.
3929 Aspects are interacting with other modules by defining advices. The concept of
3930 an \emph{advice} is known from both aspect-oriented and functional
3931 programming\citing{wikiAdvice2014}. It is a function that modifies another
3932 function when the latter is run. An advice in AspectJ is somewhat similar to a
3933 method in Java. It is meant to alter the behavior of other methods, and contains
3934 a body that is executed when it is applied.
3936 An advice can be applied at a defined \emph{pointcut}. A pointcut picks out one
3937 or more \emph{join points}. A join point is a well-defined point in the
3938 execution of a program. It can occur when calling a method defined for a
3939 particular class, when calling all methods with the same name,
3940 accessing/assigning to a particular field of a given class and so on. An advice
3941 can be declared to run both before, after returning from a pointcut, when there
3942 is thrown an exception in the pointcut or after the pointcut either returns or
3943 throws an exception. In addition to picking out join points, a pointcut can
3944 also bind variables from its context, so they can be accessed in the body of an
3945 advice. An example of a pointcut and an advice is found in
3946 \myref{lst:aspectjExample}.
3949 \begin{minted}{aspectj}
3950 pointcut methodAnalyze(
3951 SearchBasedExtractAndMoveMethodAnalyzer analyzer) :
3952 call(* SearchBasedExtractAndMoveMethodAnalyzer.analyze())
3953 && target(analyzer);
3955 after(SearchBasedExtractAndMoveMethodAnalyzer analyzer) :
3956 methodAnalyze(analyzer) {
3957 statistics.methodCount++;
3958 debugPrintMethodAnalysisProgress(analyzer.method);
3961 \caption{An example of a pointcut named \method{methodAnalyze},
3962 and an advice defined to be applied after it has occurred.}
3963 \label{lst:aspectjExample}
3966 \subsection{The Statistics class}
3967 The statistics aspect stores statistical information in an object of type
3968 \type{Statistics}. As of now, the aspect needs to be initialized at the point in
3969 time where it is desired that it starts its data gathering. At any point in time
3970 the statistics aspect can be queried for a snapshot of the current statistics.
3972 The \type{Statistics} class also includes functionality for generating a report
3973 of its gathered statistics. The report can be given either as a string or it can
3974 be written to a file.
3976 \subsection{Advices}
3977 The statistics aspect contains advices for gathering statistical data from
3978 different parts of the benchmarking process. It captures statistics from both
3979 the analysis part and the execution part of the composite \ExtractAndMoveMethod
3982 For the analysis part, there are advices to count the number of text selections
3983 analyzed and the number of methods, types, compilation units and packages
3984 analyzed. There are also advices that counts for how many of the methods there
3985 are found a selection that is a candidate for the refactoring, and for how many
3986 methods there are not.
3988 There exist advices for counting both the successful and unsuccessful executions
3989 of all the refactorings. Both for the \ExtractMethod and \MoveMethod
3990 refactorings in isolation, as well as for the combination of them.
3992 \section{Optimizations}
3993 When looking for possible optimizations for the benchmarking process, I used the
3994 \name{VisualVM}\footnote{\url{http://visualvm.java.net/}} \gloss{profiler} for
3995 the Java Virtual Machine to both profile the application and also to make memory
3998 \subsection{Caching}
3999 When \gloss{profiling} the benchmark process before making any optimizations, it
4000 early became apparent that the parsing of source code was a place to direct
4001 attention towards. This discovery was done when only \emph{analyzing} source
4002 code, before trying to do any \emph{manipulation} of it. Caching of the parsed
4003 ASTs seemed like the best way to save some time, as expected. With only a simple
4004 cache of the most recently used AST, the analysis time was speeded up by a
4005 factor of around 20. This number depends a little upon which type of system the
4008 The caching is managed by a cache manager, that now, by default, utilizes the
4009 not so well known feature of Java called a \emph{soft reference}. Soft
4010 references are best explained in the context of weak references. A \emph{weak
4011 reference} is a reference to an object instance that is only guaranteed to
4012 persist as long as there is a \emph{strong reference} or a soft reference
4013 referring the same object. If no such reference is found, its referred object is
4014 garbage collected. A strong reference is basically the same as a regular Java
4015 reference. A soft reference has the same guarantees as a week reference when it
4016 comes to its relation to strong references, but it is not necessarily garbage
4017 collected if there are no strong references to it. A soft reference \emph{may}
4018 reside in memory as long as the JVM has enough free memory in the heap. A soft
4019 reference will therefore usually perform better than a weak reference when used
4020 for simple caching and similar tasks. The way to use a soft/weak reference is to
4021 as it for its referent. The return value then has to be tested to check that it
4022 is not \var{null}. For the basic usage of soft references, see
4023 \myref{lst:softReferenceExample}. For a more thorough explanation of weak
4024 references in general, see\citing{weakRef2006}.
4027 \begin{minted}{java}
4029 Object strongRef = new Object();
4032 SoftReference<Object> softRef =
4033 new SoftReference<Object>(new Object());
4035 // Using the soft reference
4036 Object obj = softRef.get();
4041 \caption{Showing the basic usage of soft references. Weak references is used the
4042 same way. {\footnotesize (The references are part of the \code{java.lang.ref}
4044 \label{lst:softReferenceExample}
4047 The cache based on soft references has no limit for how many ASTs it caches. It
4048 is generally not advisable to keep references to ASTs for prolonged periods of
4049 time, since they are expensive structures to hold on to. For regular plugin
4050 development, \name{Eclipse} recommends not creating more than one AST at a time to
4051 limit memory consumption. Since the benchmarking has nothing to do with user
4052 experience, and throughput is everything, these advices are intentionally
4053 ignored. This means that during the benchmarking process, the target \name{Eclipse}
4054 application may very well work close to its memory limit for the heap space for
4055 long periods during the benchmark.
4057 \subsection{Candidates stored as mementos}
4058 When performing large scale analysis of source code for finding candidates to
4059 the \ExtractAndMoveMethod refactoring, memory is an issue. One of the inputs to
4060 the refactoring is a variable binding. This variable binding indirectly retains
4061 a whole AST. Since ASTs are large structures, this quickly leads to an
4062 \type{OutOfMemoryError} if trying to analyze a large project without optimizing
4063 how we store the candidates' data. This means that the JVM cannot allocate more
4064 memory for our benchmark, and it exits disgracefully.
4066 A possible solution could be to just allow the JVM to allocate even more memory,
4067 but this is not a dependable solution. The allocated memory could easily
4068 supersede the physical memory of a machine, which would make the benchmark go
4071 Thus, the candidates' data must be stored in another format. Therefore, we use
4072 the \gloss{mementoPattern} to store variable binding information. This is done
4073 in a way that makes it possible to retrieve a variable binding at a later point.
4074 The data that is stored to achieve this, is the key to the original variable
4075 binding. In addition to the key, we know which method and text selection the
4076 variable is referenced in, so that we can find it by parsing the source code and
4077 search for it when it is needed.
4079 \section{Handling failures}
4083 \chapter{Case studies}
4085 In this chapter I will present a two case studies. This is done to give an
4086 impression of how the search-based \ExtractAndMoveMethod refactoring performs
4087 when giving it a larger project to take on. I will try to answer where it lacks,
4088 in terms of completeness, as well as showing its effect on refactored source
4091 The first and primary case, is refactoring source code from the \name{Eclipse
4092 JDT UI} project. The project is chosen because it is a well-known open-source
4093 project, still in development, with a large code base that is written by many
4094 different people over several years. The code is installed in a large number of
4095 \name{Eclipse} applications worldwide, and many other projects build on the
4096 Eclipse platform. For a long time, it was even the official IDE for Android
4097 development. All this means that Eclipse must be seen as a good representative
4098 for professionally written Java source code. It is also the home for most of the
4099 JDT refactoring code.
4101 For the second case, the \ExtractAndMoveMethod refactoring is fed the
4102 \code{no.uio.ifi.refaktor} project. This is done as a variation of the
4103 ``dogfooding'' methodology.
4106 For conducting these experiments, three software tools are used. Two of the
4107 tools both use Eclipse as their platform. The first is our own tool, described
4108 in \myref{sec:benchmarking}, written to be able to run the \ExtractAndMoveMethod
4109 refactoring as a batch process. It analyzes and refactors all the methods of a
4110 project in sequence. The second is JUnit, which is used for running the
4111 project's own unit tests on the target code both before and after it is
4112 refactored. The last tool that is used is a code quality management tool, called
4113 \name{SonarQube}. It can be used to perform different tasks for assuring code
4114 quality, but we are only going to take advantage of one of its main features,
4115 namely quality profiles.
4117 A quality profile is used to define a set of coding rules that a project is
4118 supposed to comply with. Failure to following these rules will be recorded as
4119 so-called ``issues'', marked as having one of several degrees of severities,
4120 ranging from ``info'' to ``blocker'', where the latter one is the most severe.
4121 The measurements done for these case studies are therefore not presented as
4122 fine-grained software metrics results, but rather as the number of issues for
4125 In its analysis, \name{SonarQube} discriminates between functions and accessors.
4126 Accessors are methods that are recognized as setters or getters.
4128 In addition to the coding rules defined through quality profiles,
4129 \name{SonarQube} calculates the complexity of source code. The metric that is
4130 used is cyclomatic complexity, developed by Thomas J. McCabe in
4131 1976\citing{mccabeComplexity1976}. In this metric, functions have an initial
4132 complexity of 1, and whenever the control flow of a function splits, the
4133 complexity increases by
4134 one\footnote{\url{http://docs.codehaus.org/display/SONAR/Metric+definitions}}.
4135 Accessors are not counted in the complexity analysis.
4137 Specifications for the computer used during the experiments are shown in
4138 \myref{tab:experimentComputerSpecs}.
4141 \caption{Specifications for experiment computer.}
4142 \label{tab:experimentComputerSpecs}
4144 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{0.35}R{1.65}@{}}
4146 \spancols{2}{Hardware} \\
4148 Model & Lenovo ThinkPad Edge S430 \\
4149 Processor & Intel\textregistered{} Core\texttrademark{}
4150 i5-3210M\linebreak[4] (2.5 GHz/3.1 GHz (turbo),
4151 2 cores, 4 threads, 3 MB Cache) \\
4152 Memory & 8 GB DDR3 1600 MHz \\
4153 Storage & 500 GB HDD (7200 RPM) + 16 GB SSD Cache for Lenovo Hard Disk Drive
4154 Performance Booster \\
4156 \spancols{2}{Operating system} \\
4158 Distribution & Ubuntu 12.10 \\
4159 Kernel & Linux 3.5.0-49-generic (x86\_64) \\
4166 \section{The \name{SonarQube} quality profile}
4167 The quality profile that is used with \name{SonarQube} in these case studies has got
4168 the name \name{IFI Refaktor Case Study} (version 6). The rules defined in the
4169 profile are chosen because they are the available rules found in \name{SonarQube} that
4170 measures complexity and coupling. Now follows a description of the rules in the
4171 quality profile. The values that are set for these rules are listed in
4172 \myref{tab:qualityProfile1}.
4175 \item[Avoid too complex class] is a rule that measures cyclomatic complexity
4176 for every statement in the body of a class, except for setters and getter.
4177 The threshold value set is its default value of 200.
4179 \item[Classes should not be coupled to too many other classes ] is a rule that
4180 measures how many other classes a class depends upon. It does not count the
4181 dependencies of nested classes. It is meant to promote the Single
4182 Responsibility Principle. The metric for the rule resembles the CBO metric
4183 that is described in \myref{sec:CBO}, but is only considering outgoing
4184 dependencies. The max value for the rule is chosen on the basis of an
4185 empirical study by Raed Shatnawi, which concludes that the number 9 is the
4186 most useful threshold for the CBO metric\citing{shatnawiQuantitative2010}.
4187 This study is also performed on Eclipse source code, so this threshold value
4188 should be particularly well suited for the Eclipse JDT UI case in this
4191 \item[Control flow statements \ldots{} should not be nested too deeply] is
4192 a rule that is meant to counter ``Spaghetti code''. It measures the nesting
4193 level of \emph{if}, \emph{for}, \emph{while}, \emph{switch} and \emph{try}
4194 statements. The nesting levels start at 1. The max value set is its default
4197 \item[Methods should not be too complex] is a rule that measures cyclomatic
4198 complexity the same way as the ``Avoid too complex class'' rule. The max
4199 value used is 10, which ``seems like a reasonable, but not magical, upper
4200 limit``\citing{mccabeComplexity1976}.
4202 \item[Methods should not have too many lines] is a rule that simply measures
4203 the number of lines in methods. A threshold value of 20 is used for this
4204 metric. This is based on my own subjective opinions, as the default value of
4205 100 describes method bodies that do not even fit on most screens.
4207 \item[NPath Complexity] is a rule that measures the number of possible
4208 execution paths through a function. The value used is the default value of
4209 200, which seems like a recognized threshold for this metric.
4211 \item[Too many methods] is a rule that measures the number of methods in a
4212 class. The threshold value used is the default value of 10.
4218 \caption{The \name{IFI Refaktor Case Study} quality profile (version 6).}
4219 \label{tab:qualityProfile1}
4221 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{1.5}R{0.5}@{}}
4223 \textbf{Rule} & \textbf{Max value} \\
4225 Avoid too complex class & 200 \\
4226 Classes should not be coupled to too many other classes (Single
4227 Responsibility Principle) & 9 \\
4228 Control flow statements \ldots{} should not be nested too deeply &
4230 Methods should not be too complex & 10 \\
4231 Methods should not have too many lines & 20 \\
4232 NPath Complexity & 200 \\
4233 Too many methods & 10 \\
4240 A precondition for the source code that is going to be the target for a series
4241 of \ExtractAndMoveMethod refactorings, is that it is organized as an Eclipse
4242 project. It is also assumed that the code is free from compilation errors.
4244 \section{The experiment}
4245 For a given project, the first job that is done, is to refactor its source code.
4246 The refactoring batch job produces three things: The refactored project,
4247 statistics gathered during the execution of the series of refactorings, and an
4248 error log describing any errors happening during this execution. See
4249 \myref{sec:benchmarking} for more information about how the refactorings are
4252 After the refactoring process is done, the before- and after-code is analyzed
4253 with \name{SonarQube}. The analysis results are then stored in a database and
4254 displayed through a \name{SonarQube} server with a web interface.
4256 The before- and after-code is also tested with their own unit tests. This is
4257 done to discover any changes in the semantic behavior of the refactored code,
4258 within the limits of these tests.
4260 \section{Case 1: The Eclipse JDT UI project}
4261 This case is the ultimate test for our \ExtractAndMoveMethod refactoring. The
4262 target source code is massive. With its over 300,000 lines of code\footnote{For
4263 all uses of ``lines of code'' we follow the definition from \name{SonarQube}.
4264 LOC = the number of physical lines containing a character which is neither
4265 whitespace or part of a comment.} and over 25,000 methods, it is a formidable
4266 task to perform automated changes on it. There should be plenty of situations
4267 where things can go wrong, and, as we shall see later, they do.
4269 I will start by presenting some statistics from the refactoring execution,
4270 before I pick apart the \name{SonarQube} analysis and conclude by commenting on
4271 the results from the unit tests. The configuration for the experiment is
4272 specified in \myref{tab:configurationCase1}.
4275 \caption{Configuration for Case 1.}
4276 \label{tab:configurationCase1}
4278 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{0.67}L{1.33}@{}}
4280 \spancols{2}{Benchmark data} \\
4282 Launch configuration & CaseStudy.launch \\
4283 Project & no.uio.ifi.refaktor.benchmark \\
4284 Repository & gitolite@git.uio.no:ifi-stolz-refaktor \\
4285 Commit & 43c16c04520746edd75f8dc2a1935781d3d9de6c \\
4287 \spancols{2}{Input data} \\
4289 Project & org.eclipse.jdt.ui \\
4290 Repository & git://git.eclipse.org/gitroot/jdt/eclipse.jdt.ui.git \\
4291 Commit & f218388fea6d4ec1da7ce22432726c244888bb6b \\
4292 Branch & R3\_8\_maintenance \\
4293 Tests suites & org.eclipse.jdt.ui.tests.AutomatedSuite,
4294 org.eclipse.jdt.ui.tests.refactoring.all.\-AllAllRefactoringTests \\
4299 \subsection{Statistics}
4300 The statistics gathered during the refactoring execution is presented in
4301 \myref{tab:case1Statistics}.
4304 \caption{Statistics after batch refactoring the Eclipse JDT UI project with
4305 the \ExtractAndMoveMethod refactoring.}
4306 \label{tab:case1Statistics}
4308 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{1.5}R{0.5}@{}}
4310 \spancols{2}{Time used} \\
4312 Total time & 98m38s \\
4313 Analysis time & 14m41s (15\%) \\
4314 Change time & 74m20s (75\%) \\
4315 Miscellaneous tasks & 9m37s (10\%) \\
4317 \spancols{2}{Numbers of each type of entity analyzed} \\
4320 Compilation units & 2,097 \\
4323 Text selections & 591,500 \\
4325 \spancols{2}{Numbers for \ExtractAndMoveMethod refactoring candidates} \\
4327 Methods chosen as candidates & 2,552 \\
4328 Methods NOT chosen as candidates & 25,115 \\
4329 Candidate selections (multiple per method) & 36,843 \\
4331 \spancols{2}{\ExtractAndMoveMethod refactorings executed} \\
4333 Fully executed & 2,469 \\
4334 Not fully executed & 83 \\
4335 Total attempts & 2,552 \\
4337 \spancols{2}{Primitive refactorings executed} \\
4338 \spancols{2}{\small \ExtractMethod refactorings} \\
4340 Performed & 2,483 \\
4341 Not performed & 69 \\
4342 Total attempts & 2,552 \\
4344 \spancols{2}{\small \MoveMethod refactorings} \\
4347 Not performed & 14 \\
4348 Total attempts & 2,483 \\
4354 \subsubsection{Execution time}\label{sec:case1ExecutionTime}
4355 I consider the total execution time of approximately 1.5 hours, on a mid-level
4356 laptop computer, as being acceptable. It clearly makes the batch process
4357 unsuitable for doing any on-demand analysis or changes, but it is good enough
4358 for running periodic jobs, like over-night analysis. In comparison, the
4359 SonarQube analysis for the same project consumes about the same amount of time.
4361 As the statistics show, 75\% of the total time goes into making the actual code
4362 changes. The time consumers are here the primitive \ExtractMethod and
4363 \MoveMethod refactorings. Included in the change time is the parsing and
4364 precondition checking done by the refactorings, as well as textual changes done
4365 to files on disk. All this parsing and disk access is time-consuming, and
4366 constitutes a large part of the change time.
4368 The pure analysis time, which is the time used on finding suitable refactoring
4369 candidates, only makes up for 15\% of the total time consumed. This includes
4370 analyzing almost 600,000 text selections, while the number of attempted
4371 executions of the \ExtractAndMoveMethod refactoring is only about 2,500. So the
4372 number of executed primitive refactorings is approximately 5,000. Assuming the
4373 time used on miscellaneous tasks are used mostly for parsing source code for the
4374 analysis, we can say that the time used for analyzing code is at most 25\% of
4375 the total time. This means that for every primitive refactoring executed, we
4376 can analyze about 360 text selections. So, with an average of about 21 text
4377 selections per method, it is reasonable to say that we can analyze over 15
4378 methods in the time it takes to perform a primitive refactoring.
4380 \subsubsection{Refactoring candidates}
4381 Out of the 27,667 methods that were analyzed, 2,552 methods contained selections
4382 that were considered candidates for the \ExtractAndMoveMethod refactoring. This
4383 is roughly 9\% off the methods in the project. These 9\% of the methods had on
4384 average 14.4 text selections that were considered possible refactoring
4387 \subsubsection{Executed refactorings}
4388 2,469 out of 2,552 attempts on executing the \ExtractAndMoveMethod refactoring
4389 were successful, giving a success rate of 96.7\%. The failure rate of 3.3\%
4390 stems from situations where the analysis finds a candidate selection, but the
4391 change execution fails. This failure could be an exception that was thrown, and
4392 the refactoring aborts. It could also be the precondition checking for one of
4393 the primitive refactorings that gives us an error status, meaning that if the
4394 refactoring proceeds, the code will contain compilation errors afterwards,
4395 forcing the composite refactoring to abort. This means that if the
4396 \ExtractMethod refactoring fails, no attempt is done for the \MoveMethod
4397 refactoring. \todo{Redundant information? Put in benchmark chapter?}
4399 Out of the 2,552 \ExtractMethod refactorings that were attempted executed, 69 of
4400 them failed. This gives a failure rate of 2.7\% for the primitive refactoring.
4401 In comparison, the \MoveMethod refactoring had a failure rate of 0.6 \% of the
4402 2,483 attempts on the refactoring.
4404 The failure rates for the refactorings are not that bad, if we also take into
4405 account that the pre-refactoring analysis is incomplete.\todo{see \ldots}
4407 \subsection{\name{SonarQube} analysis}
4408 Results from the \name{SonarQube} analysis are shown in
4409 \myref{tab:case1ResultsProfile1}.
4412 \caption{Results for analyzing the Eclipse JDT UI project, before and after
4413 the refactoring, with \name{SonarQube} and the \name{IFI Refaktor Case Study}
4414 quality profile. (Bold numbers are better.)}
4415 \label{tab:case1ResultsProfile1}
4417 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{1.5}R{0.25}R{0.25}@{}}
4419 \textnormal{Number of issues for each rule} & Before & After \\
4421 Avoid too complex class & 81 & \textbf{79} \\
4422 Classes should not be coupled to too many other classes (Single
4423 Responsibility Principle) & \textbf{1,098} & 1,199 \\
4424 Control flow statements \ldots{} should not be nested too deeply & 1,375 &
4426 Methods should not be too complex & 1,518 & \textbf{1,452} \\
4427 Methods should not have too many lines & 3,396 & \textbf{3,291} \\
4428 NPath Complexity & 348 & \textbf{329} \\
4429 Too many methods & \textbf{454} & 520 \\
4431 Total number of issues & 8,270 & \textbf{8,155} \\
4434 \spancols{3}{Complexity} \\
4436 Per function & 3.6 & \textbf{3.3} \\
4437 Per class & \textbf{29.5} & 30.4 \\
4438 Per file & \textbf{44.0} & 45.3 \\
4440 Total complexity & \textbf{84,765} & 87,257 \\
4443 \spancols{3}{Numbers of each type of entity analyzed} \\
4445 Files & 1,926 & 1,926 \\
4446 Classes & 2,875 & 2,875 \\
4447 Functions & 23,744 & 26,332 \\
4448 Accessors & 1,296 & 1,019 \\
4449 Statements & 162,768 & 165,145 \\
4450 Lines of code & 320,941 & 329,112 \\
4452 Technical debt (in days) & \textbf{1,003.4} & 1,032.7 \\
4457 \subsubsection{Diversity in the number of entities analyzed}
4458 The analysis performed by \name{SonarQube} is reporting fewer methods than found
4459 by the pre-refactoring analysis. \name{SonarQube} discriminates between
4460 functions (methods) and accessors, so the 1,296 accessors play a part in this
4461 calculation. \name{SonarQube} also has the same definition as our plugin when
4462 it comes to how a class is defined. Therefore it seems like \name{SonarQube}
4463 misses 277 classes that our plugin handles. This can explain why the {SonarQube}
4464 report differs from our numbers by approximately 2,500 methods.
4466 \subsubsection{Complexity}
4467 On all complexity rules that works on the method level, the number of issues
4468 decreases with between 3.1\% and 6.5\% from before to after the refactoring. The
4469 average complexity of a method decreases from 3.6 to 3.3, which is an
4470 improvement of about 8.3\%. So, on the method level, the refactoring must be
4471 said to have a slightly positive impact. This is due to the extraction of a lot
4472 of methods, making the average method size smaller.
4474 The improvement in complexity on the method level is somewhat traded for
4475 complexity on the class level. The complexity per class metric is worsened by
4476 3\% from before to after. The issues for the ``Too many methods'' rule also
4477 increases by 14.5\%. These numbers indicate that the refactoring makes quite a
4478 lot of the classes a little more complex overall. This is the expected outcome,
4479 since the \ExtractAndMoveMethod refactoring introduces almost 2,500 new methods
4482 The only number that can save the refactoring's impact on complexity on the
4483 class level, is the ``Avoid too complex class'' rule. It improves with 2.5\%,
4484 thus indicating that the complexity is moderately better distributed between the
4485 classes after the refactoring than before.
4487 \subsubsection{Coupling}
4488 One of the hopes when starting this project, was to be able to make a
4489 refactoring that could lower the coupling between classes. Better complexity at
4490 the method level is a not very unexpected byproduct of dividing methods into
4491 smaller parts. Lowering the coupling on the other hand, is a far greater task.
4492 This is also reflected in the results for the only coupling rule defined in the
4493 \name{SonarQube} quality profile, namely the ``Classes should not be coupled to
4495 other classes (Single Responsibility Principle)'' rule.
4497 The number of issues for the coupling rule is 1,098 before the refactoring, and
4498 1,199 afterwards. This is an increase in issues of 9.2\%. These numbers can be
4499 interpreted two ways. The first possibility is that our assumptions are wrong,
4500 and that increasing indirection does not decrease coupling between classes. The
4501 other possibility is that our analysis and choices of candidate text selections
4502 are not good enough. I vote for the second possibility. (Voting against the
4503 public opinion may also be a little bold.)
4505 \subsubsection{An example of what makes the number of dependency issues grow}
4506 \Myref{lst:sonarJDTExampleBefore} shows a portion of the class
4507 \typewithref{org.eclipse.jdt.ui.actions}{ShowActionGroup} from the JDT UI
4508 project before it is refactored with the search-based \ExtractAndMoveMethod
4509 refactoring. Before the refactoring, the \type{ShowActionGroup} class has 12
4510 outgoing dependencies (reported by \name{SonarQube}).
4512 \begin{listing}[htb]
4513 \begin{minted}[linenos,samepage]{java}
4514 public class ShowActionGroup extends ActionGroup {
4516 private void initialize(IWorkbenchSite site,
4517 boolean isJavaEditor) {
4519 ISelectionProvider provider= fSite.getSelectionProvider();
4520 ISelection selection= provider.getSelection();
4521 fShowInPackagesViewAction.update(selection);
4522 if (!isJavaEditor) {
4523 provider.addSelectionChangedListener(
4524 fShowInPackagesViewAction);
4529 \caption{Portion of the \type{ShowActionGroup} class before refactoring.}
4530 \label{lst:sonarJDTExampleBefore}
4533 During the benchmark process, the search-based \ExtractAndMoveMethod refactoring
4534 extracts the lines 6 to 12 of the code in \myref{lst:sonarJDTExampleBefore}, and
4535 moves the new method to the move target, which is the field
4536 \var{fShowInPackagesViewAction} with type
4537 \typewithref{org.eclipse.jdt.ui.actions}{ShowInPackageViewAction}. The result is
4538 shown in \myref{lst:sonarJDTExampleAfter}.
4540 \begin{listing}[htb]
4541 \begin{minted}[linenos,samepage]{java}
4542 public class ShowActionGroup extends ActionGroup {
4544 private void initialize(IWorkbenchSite site,
4545 boolean isJavaEditor) {
4547 fShowInPackagesViewAction.generated_8019497110545412081(
4548 this, isJavaEditor);
4553 \begin{minted}[linenos,samepage]{java}
4554 public class ShowInPackageViewAction
4555 extends SelectionDispatchAction {
4557 public void generated_8019497110545412081(
4558 ShowActionGroup showactiongroup, boolean isJavaEditor) {
4559 ISelectionProvider provider=
4560 showactiongroup.fSite.getSelectionProvider();
4561 ISelection selection= provider.getSelection();
4563 if (!isJavaEditor) {
4564 provider.addSelectionChangedListener(this);
4569 \caption{Portions of the classes \type{ShowActionGroup} and
4570 \type{ShowInPackageViewAction} after refactoring.}
4571 \label{lst:sonarJDTExampleAfter}
4574 After the refactoring, the \type{ShowActionGroup} has only 11 outgoing
4575 dependencies. It no longer depends on the
4576 \typewithref{org.eclipse.jface.viewers}{ISelection} type. So our refactoring
4577 managed to get rid of one dependency, which is exactly what we wanted. The only
4578 problem is, that now the \type{ShowInPackageViewAction} class has got two new
4579 dependencies, in the \type{ISelectionProvider} and the \type{ISelection} types.
4580 The bottom line is that we eliminated one dependency, but introduced two more,
4581 ending up with a program that has more dependencies now than when we started.
4583 What can happen in many situations where the \ExtractAndMoveMethod refactoring
4584 is performed, is that the \MoveMethod refactoring ``drags'' with it references
4585 to classes that are unknown to the method destination. If the refactoring
4586 happens to be so lucky that it removes a dependency from one class, it might as
4587 well introduce a couple of new dependencies to another class, as shown in the
4588 previous example. In those situations where a destination class does not know
4589 about the originating class of a moved method, the \MoveMethod refactoring most
4590 certainly will introduce a dependency. This is because there is a
4591 bug\footnote{\href{https://bugs.eclipse.org/bugs/show\_bug.cgi?id=228635}{Eclipse
4592 Bug 228635 - [move method] unnecessary reference to source}} in the refactoring,
4593 making it pass an instance of the originating class as a reference to the moved
4594 method, regardless of whether the reference is used in the method body or not.
4596 There is also the possibility that the heuristics used to find candidate text
4597 selections are not good enough. There is work to be done with fine-tuning the
4598 heuristics and to complete the analysis part of this project.
4600 \subsubsection{Totals}
4601 On the bright side, the total number of issues is lower after the refactoring
4602 than it was before. Before the refactoring, the total number of issues was
4603 8,270, and after it is 8,155. This is an improvement of 1.4\%.
4605 The down side is that \name{SonarQube} shows that the total cyclomatic
4606 complexity has increased by 2.9\%, and that the (more questionable) ``technical
4607 debt'' has increased from 1,003.4 to 1,032.7 days, also a deterioration of
4608 2.9\%. Although these numbers are similar, no correlation has been found
4611 \subsection{Unit tests}
4612 The tests that have been run for the \name{Eclipse JDT UI} project, are the
4613 test suites specified as the main test suites on the JDT UI wiki page on how to
4615 project\footnote{\url{https://wiki.eclipse.org/JDT\_UI/How\_to\_Contribute\#Unit\_Testing}}.
4616 The results from these tests are shown in \myref{tab:case1UnitTests}.
4619 \caption{Results from the unit tests run for the Eclipse JDT UI project,
4620 before and after the refactoring.}
4621 \label{tab:case1UnitTests}
4623 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{1}R{0.5}R{0.5}@{}}
4625 \textnormal{AutomatedSuite} & Before & After \\
4627 Runs & 2007/2007 & 2007/2007 \\
4631 \spancols{2}{AllAllRefactoringTests} \\
4633 Runs & 3815/3816 & 3815/3816 \\
4634 Errors & 2 & 2257 \\
4640 \subsubsection{Before the refactoring}
4641 Running the tests for the before-code of Eclipse JDT UI yielded 4 errors and 3
4642 failures for the \type{AutomatedSuite} test suite (2,007 test cases), and 2
4643 errors and 3 failures for the \type{AllAllRefactoringTests} test suite (3,816
4646 \subsubsection{After the refactoring}
4647 For the after-code of the Eclipse JDT UI project, Eclipse reports that the
4648 project contains 322 compilation errors, and a lot of other errors that
4649 follow from these. All of the errors are caused by the \ExtractAndMoveMethod
4650 refactoring. Had these errors originated from only one bug, it would not have
4651 been much of a problem, but this is not the case. By only looking at some random
4652 compilation problems in the refactored code, I came up with at least four
4653 different bugs \todo{write bug reports?} that caused those problems. I then
4654 stopped looking for more, since some of the bugs would take more time to fix
4655 than I could justify using on them at this point.
4657 One thing that can be said in my defense, is that all the compilation errors
4658 could have been avoided if the types of situations that cause them were properly
4659 handled by the primitive refactorings, which again are supplied by the Eclipse
4660 JDT UI project. All four bugs that I mentioned before are weaknesses of the
4661 \MoveMethod refactoring. If the primitive refactorings had detected the
4662 up-coming errors in their precondition checking phase, the refactorings would
4663 have been aborted, since this is how the \ExtractAndMoveMethod refactoring
4664 handles such situations. This shows that it is not safe to completely rely upon
4665 the primitive refactorings to save us if our own pre-refactoring analysis fails
4666 to detect that a compilation error will be introduced. A problem is that the
4667 source code analysis done by both the JDT refactorings and our own tool is
4670 Of course, taking into account all possible situations that could lead to
4671 compilation errors is an immense task. A complete analysis of these situations
4672 is too big of a problem for this master's project to solve. Looking at it now,
4673 this comes as no surprise, since the task is obviously also too big for the
4674 creators of the primitive \MoveMethod refactoring.
4676 Considering all these problems, it is difficult to know how to interpret the
4677 unit test results from after refactoring the Eclipse JDT UI. The
4678 \type{AutomatedSuite} reported 565 errors and 5 failures, which means that 1437,
4679 or 71.6\%, of the tests still passed. Three of the failures were the same as
4680 reported before the refactoring took place, so two of them are new. For these
4681 two cases it is not immediately apparent what makes them behave differently. The
4682 program is so complex that to analyze it to find this out, we might need more
4683 powerful methods than just manually analyzing its source code. This is somewhat
4684 characteristic for imperative programming: The programs are often hard to
4685 analyze and understand.
4687 For the \type{AllAllRefactoringTests} test suite, the three failures are gone,
4688 but the two errors have grown to 2,257 errors. I will not try to analyze those
4691 What I can say at this point, is that it is likely that the
4692 \ExtractAndMoveMethod refactoring has introduced some unintentional behavioral
4693 changes. Let us say that the refactoring introduces at least two
4694 behavior-altering changes for every 2,500 executions. More than that is
4695 difficult to say about the behavior-preserving properties of the
4696 \ExtractAndMoveMethod refactoring, at this point.
4698 \subsection{Conclusions}
4699 After automatically analyzing and executing the \ExtractAndMoveMethod
4700 refactoring for all the methods in the Eclipse JDT UI project, the results do
4701 not look that promising. For this case, the refactoring seems almost unusable as
4702 it is now. The error rate and measurements tell us this.
4704 The refactoring makes the code a little less complex at the method level. But
4705 this is merely a side effect of extracting methods. When it comes to the overall
4706 complexity, it is increased, although it is slightly better spread among the
4709 The analysis done before the \ExtractAndMoveMethod refactoring, is currently not
4710 complete enough to make the refactoring useful. It introduces too many errors in
4711 the code, and the code may change its behavior. It also remains to prove that
4712 large scale refactoring with it can decrease coupling between classes. A better
4713 analysis may prove this, but in its present state, the opposite is the fact. The
4714 coupling measurements done by \name{SonarQube} show this.
4716 On the bright side, the performance of the refactoring process is not that bad.
4717 It shows that it is possible to make a tool the way we do, if we can make the
4718 tool do anything useful. As long as the analysis phase is not going to involve
4719 anything that uses too much disk access, a lot of analysis can be done in a
4720 reasonable amount of time.
4722 The time used on performing the actual changes excludes a trial and error
4723 approach with the tools used in this master's project. In a trial and error
4724 approach, you could for instance be using the primitive refactorings used in
4725 this project to refactor code, and only then make decisions based on the effect,
4726 possibly shown by traditional software metrics. The problem with the approach
4727 taken in this project, compared to a trial and error approach, is that using
4728 heuristics beforehand is much more complicated. But on the other hand, a trial
4729 and error approach would still need to face the challenges of producing code
4730 that does compile without errors. If using refactorings that could produce
4731 in-memory changes, a trial and error approach could be made more efficient.
4733 \section{Case 2: The \type{no.uio.ifi.refaktor} project}
4734 In this case we will see a form of the ``dogfooding'' methodology used, when
4735 refactoring our own \type{no.uio.ifi.refaktor} project with the
4736 \ExtractAndMoveMethod refactoring.
4738 In this case I will try to point out some differences from the first case, and
4739 how they impact the execution of the benchmark. The refaktor project is 39 times
4740 smaller than the Eclipse JDT UI project, measured in lines of code. This will
4741 make things a bit more transparent. It will therefore be interesting to see if
4742 this case can shed light on any aspect of our project that were lost in the
4745 The configuration for the experiment is specified in
4746 \myref{tab:configurationCase2}.
4749 \caption{Configuration for Case 2.}
4750 \label{tab:configurationCase2}
4752 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{0.67}L{1.33}@{}}
4754 \spancols{2}{Benchmark data} \\
4756 Launch configuration & CaseStudyDogfooding.launch \\
4757 Project & no.uio.ifi.refaktor.benchmark \\
4758 Repository & gitolite@git.uio.no:ifi-stolz-refaktor \\
4759 Commit & 43c16c04520746edd75f8dc2a1935781d3d9de6c \\
4761 \spancols{2}{Input data} \\
4763 Project & no.uio.ifi.refaktor \\
4764 Repository & gitolite@git.uio.no:ifi-stolz-refaktor \\
4765 Commit & 43c16c04520746edd75f8dc2a1935781d3d9de6c \\
4767 Test configuration & no.uio.ifi.refaktor.tests/ExtractTest.launch \\
4772 \subsection{Statistics}
4773 The statistics gathered during the refactoring execution is presented in
4774 \myref{tab:case2Statistics}.
4777 \caption{Statistics after batch refactoring the \type{no.uio.ifi.refaktor}
4778 project with the \ExtractAndMoveMethod refactoring.}
4779 \label{tab:case2Statistics}
4781 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{1.5}R{0.5}@{}}
4783 \spancols{2}{Time used} \\
4785 Total time & 1m15s \\
4786 Analysis time & 0m18s (24\%) \\
4787 Change time & 0m47s (63\%) \\
4788 Miscellaneous tasks & 0m10s (14\%) \\
4790 \spancols{2}{Numbers of each type of entity analyzed} \\
4793 Compilation units & 154 \\
4796 Text selections & 8,609 \\
4798 \spancols{2}{Numbers for \ExtractAndMoveMethod refactoring candidates} \\
4800 Methods chosen as candidates & 58 \\
4801 Methods NOT chosen as candidates & 1,012 \\
4802 Candidate selections (multiple per method) & 227 \\
4804 \spancols{2}{\ExtractAndMoveMethod refactorings executed} \\
4806 Fully executed & 53 \\
4807 Not fully executed & 5 \\
4808 Total attempts & 58 \\
4810 \spancols{2}{Primitive refactorings executed} \\
4811 \spancols{2}{\small \ExtractMethod refactorings} \\
4814 Not performed & 2 \\
4815 Total attempts & 58 \\
4817 \spancols{2}{\small \MoveMethod refactorings} \\
4820 Not performed & 3 \\
4821 Total attempts & 56 \\
4827 \subsubsection{Differences}
4828 There are some differences between the two projects that make them a little
4829 difficult to compare by performance.
4831 \paragraph{Different complexity.}
4832 Although the JDT UI project is 39 times greater than the refaktor project in
4833 terms of lines of code, it is only about 26 times its size measured in numbers
4834 of methods. This means that the methods in the refaktor project are smaller in
4835 average than in the JDT project. This is also reflected in the \name{SonarQube}
4836 report, where the complexity per method for the JDT project is 3.6, while the
4837 refaktor project has a complexity per method of 2.1.
4839 \paragraph{Number of selections per method.}
4840 The analysis for the JDT project processed 21 text selections per method in
4841 average. This number for the refaktor project is only 8 selections per method
4842 analyzed. This is a direct consequence of smaller methods.
4844 \paragraph{Different candidates to methods ratio.}
4845 The differences in how the projects are factored are also reflected in the
4846 ratios for how many methods that are chosen as candidates compared to the total
4847 number of methods analyzed. For the JDT project, 9\% of the methods were
4848 considered to be candidates, while for the refaktor project, only 5\% of the
4849 methods were chosen.
4851 \paragraph{The average number of possible candidate selection.}
4852 For the methods that are chosen as candidates, the average number of possible
4853 candidate selections for these methods differ quite much. For the JDT project,
4854 the number of possible candidate selections for these methods was 14.44
4855 selections per method, while the candidate methods in the refaktor project had
4856 only 3.91 candidate selections to choose from, in average.
4858 \subsubsection{Execution time}
4859 The differences in complexity, and the different candidate methods to total
4860 number of methods ratios, is shown in the distributions of the execution times.
4861 For the JDT project, 75\% of the total time was used on the actual changes,
4862 while for the refaktor project, this number was only 63\%.
4864 For the JDT project, the benchmark used on average 0.21 seconds per method in
4865 the project, while for the refaktor project it used only 0.07 seconds per
4866 method. So the process used 3 times as much time per method for the JDT project
4867 than for the refaktor project.
4869 While the JDT project is 39 times larger than the refaktor project measured in
4870 lines of code, the benchmark used about 79 times as long time on it than for the
4871 refaktor project. Relatively, this is about twice as long.
4873 Since the details of these execution times are not that relevant to this
4874 master's project, only their magnitude, I will leave them here.
4876 \subsubsection{Executed refactorings}
4877 For the composite \ExtractAndMoveMethod refactoring performed in case 2, 53
4878 successful attempts out of 58 gives a success rate of 91.4\%. This is 5.3
4879 percentage points worse than for the first case.
4881 \subsection{\name{SonarQube} analysis}
4882 Results from the \name{SonarQube} analysis are shown in
4883 \myref{tab:case2ResultsProfile1}.
4885 Not much is to be said about these results. The trends in complexity and
4886 coupling are the same. We end up a little worse after the refactoring process
4890 \caption{Results for analyzing the \var{no.uio.ifi.refaktor} project, before
4891 and after the refactoring, with \name{SonarQube} and the \name{IFI Refaktor
4892 Case Study} quality profile. (Bold numbers are better.)}
4893 \label{tab:case2ResultsProfile1}
4895 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{1.5}R{0.25}R{0.25}@{}}
4897 \textnormal{Number of issues for each rule} & Before & After \\
4899 Avoid too complex class & 1 & 1 \\
4900 Classes should not be coupled to too many other classes (Single
4901 Responsibility Principle) & \textbf{29} & 34 \\
4902 Control flow statements \ldots{} should not be nested too deeply & 24 &
4904 Methods should not be too complex & 17 & \textbf{15} \\
4905 Methods should not have too many lines & 41 & \textbf{40} \\
4906 NPath Complexity & 3 & 3 \\
4907 Too many methods & \textbf{13} & 15 \\
4909 Total number of issues & \textbf{128} & 129 \\
4912 \spancols{3}{Complexity} \\
4914 Per function & 2.1 & 2.1 \\
4915 Per class & \textbf{12.5} & 12.9 \\
4916 Per file & \textbf{13.8} & 14.2 \\
4918 Total complexity & \textbf{2,089} & 2,148 \\
4921 \spancols{3}{Numbers of each type of entity analyzed} \\
4923 Files & 151 & 151 \\
4924 Classes & 167 & 167 \\
4925 Functions & 987 & 1,045 \\
4926 Accessors & 35 & 30 \\
4927 Statements & 3,355 & 3,416 \\
4928 Lines of code & 8,238 & 8,460 \\
4930 Technical debt (in days) & \textbf{19.0} & 20.7 \\
4935 \subsection{Unit tests}
4936 The tests used for this case are the same that has been developed throughout
4937 this master's project.
4939 The code that was refactored for this case suffered from some of the problems
4940 discovered in the first case. This means that the after-code for this case also
4941 contained compilation errors, but they were not as many. The code contained only
4942 6 errors that made the code not compile.
4944 All of the six errors originated from the same bug. The bug arises in a
4945 situation where a class instance creation is moved between packages, and the
4946 class for the instance is package-private. The \MoveMethod refactoring does not
4947 detect that there will be a visibility problem, and neither does it promote the
4948 package-private class to be public.
4950 Since the errors in the refactored refaktor code were easy to fix manually, I
4951 corrected them and ran the unit tests as planned. The unit test results are
4952 shown in \myref{tab:case2UnitTests}. Before the refactoring, all tests passed.
4953 All tests also passed after the refactoring, with the six error corrections.
4954 Since the corrections done are not of a kind that could make the behavior of the
4955 program change, it is likely that the refactorings done to the
4956 \type{no.uio.ifi.refaktor} project did not change its behavior. This is also
4957 supported by the informal experiment presented next.
4960 \caption{Results from the unit tests run for the \type{no.uio.ifi.refaktor}
4961 project, before and after the refactoring (with 6 corrections done to the
4963 \label{tab:case2UnitTests}
4965 \begin{tabularx}{\textwidth}{@{}>{\bfseries}L{1}R{0.5}R{0.5}@{}}
4969 Runs & 148/148 & 148/148 \\
4976 \subsection{An additional experiment}
4977 To complete the task of ``eating my own dog food'', I conducted an experiment
4978 where I used the refactored version of the \type{no.uio.ifi.refaktor} project,
4979 with the corrections, to again refaktor ``itself''.
4981 The experiment produced code containing the same six errors as after the
4982 previous experiment. I also compared the after-code from the two experiments
4983 with a diff-tool. The only differences found were different method names. This
4984 is expected, since the method names are randomly generated by the
4985 \ExtractAndMoveMethod refactoring.
4987 The outcome of this simple experiment makes me more confident that the
4988 \ExtractAndMoveMethod refactoring made only behavior-preserving changes to the
4989 \type{no.uio.ifi.refaktor} project, apart from the compilation errors.
4991 \subsection{Conclusions}
4992 The differences in complexity between the Eclipse JDT UI project and the
4993 \type{no.uio.ifi.refaktor} project, clearly influenced the differences in their
4994 execution times. This is mostly because fewer of the methods were chosen to be
4995 refactored for the refaktor project than for the JDT project. This makes it
4996 difficult to know if there are any severe performance penalties associated with
4997 refactoring on a large project compared to a small one.
4999 The trends in the \name{SonarQube} analysis are the same for this case as for
5000 the previous one. This gives more confidence in the these results.
5002 By refactoring our own code and using it again to refactor our code, we showed
5003 that it is possible to write an automated composite refactoring that works for
5004 many cases. That it probably did not alter the behavior of a smaller project
5005 shows us nothing more than that though, and might just be a coincidence.
5008 \todoin{Write? Or wrap up in final conclusions?}
5009 \todoin{``Threats to validity''}
5012 \chapter{Conclusions and future work}
5013 This chapter will conclude this master's thesis. It will try to give justified
5014 answers to the research questions posed \see{sec:researchQuestions} and present
5015 some future work that could be done to take this project to the next level.
5017 \section{Conclusions}
5018 One of the motivations for this thesis was to create a fully automated composite
5019 refactoring that could be used to make program source code better in terms of
5020 coupling between classes. Earlier, in \mysimpleref{sec:CBO}, it was shown that a
5021 composition of the \ExtractMethod and the \MoveMethod refactorings reduces the
5022 coupling between two classes in an ideal situation. The Eclipse IDE implements
5023 both these refactorings, as well as providing a framework for analyzing source
5024 code, so it was considered a suitable tool to build upon for our project.
5026 The search-based \ExtractAndMoveMethod refactoring was created by utilizing the
5027 analysis and refactoring support of Eclipse, and a small framework was built
5028 for executing large scale refactoring with it. The refactoring was set up to
5029 analyze and execute changes on the Eclipse JDT UI project. Statistics was
5030 gathered during this process and the resulting code was analyzed through
5031 SonarQube. The project's own unit tests were also performed to find out if our
5032 refactoring introduces any behavior-altering changes in the code it refactor.
5034 \paragraph{Answering the main research question.}
5035 The first and greatest challenge was to find out if the \ExtractAndMoveMethod
5036 refactoring could be automated, in all tasks ranging from analysis to executing
5037 changes. It is now confirmed that this can be done, since it has been
5038 implemented as a part of the work done for this project. It has also been shown
5039 that the refactoring can be used to refactor large code bases, through the case
5040 study done on the Eclipse JDT UI project.
5042 If we ask if using the existing Eclipse refactorings for this task is
5043 \emph{easy}, this is another question. The refactorings provided by the JDT UI
5044 project are clearly not meant to be combined in any way. The preconditions for
5045 one refactoring are not always easily retrievable after the execution of
5046 another. Also, the refactorings are all assuming that the code they shall
5047 refactor is textualized. This means that the source code must be parsed between
5048 the executions of each refactoring. Another problem with this dependency on
5049 textual changes is that you cannot make a composition of two refactorings appear
5050 as one change if their changes overlap. This will make the undo-history of the
5051 refactoring show two changes instead of one, and is not nice for usability it
5052 the refactoring would be used as an on-demand refactoring in an IDE.
5054 Apart from the problems with implementing the actual refactoring, the analysis
5055 framework is quite nicely solved in Eclipse. The AST generated when parsing
5056 source code supports using visitors to traverse it, and this works without
5059 \paragraph{Is the refactoring efficient enough?}
5060 Since we have concluded that the search-based \ExtractAndMoveMethod refactoring
5061 is not suitable for on-demand large scale refactoring, but may be put to better
5062 use as a kind of analysis tool, superb performance is not crucial. In
5063 \myref{sec:case1ExecutionTime} we conclude that the refactoring performs well
5064 enough for this purpose. If performed on demand for a single method, the
5065 performance of the \ExtractAndMoveMethod refactoring is not an issue.
5067 \paragraph{What about breaking the source code?}
5068 The case studies showed that our safety measures that rely on the precondition
5069 checking of the existing primitive refactorings are not good enough in practice.
5070 If we were going to assure that code we change compiles, we would need to
5071 consider all possible situations where the refactoring could fail and search for
5072 them in our analysis. It is an open question if this is even feasible. Our
5073 analysis is incomplete, and so is the analysis for the \ExtractMethod and the
5074 \MoveMethod refactorings.
5076 Our refactoring does not take any precautions to preserve behavior. A few
5077 running and failing unit test for the JDT UI project after the refactoring
5078 indicate that our refactoring probably causes some changes to the way a program
5081 \paragraph{Is the quality of the code improved?}
5082 For coupling, there is no evidence that the refactoring improves the quality of
5083 source code. Shall we believe the SonarQube analysis from the case studies, our
5084 refactoring makes classes more coupled after the refactoring than before, in the
5085 general case. This is probably because our analysis and heuristics for finding
5086 the best candidates for the refactoring are not adequate.
5088 \paragraph{Is the refactoring useful?}
5089 In its present state, the refactoring cannot be said to be very useful. It
5090 generates too many compilation errors for it to fall into that category. On the
5091 other hand, if the problems with the search-based \ExtractAndMoveMethod
5092 refactoring were to be solved it could be useful in some situations.
5094 If the refactoring was perfected, it could of course be used as a regular
5095 on-demand automated refactoring on a per method base (or per class, package or
5098 As it is now, the refactoring is not very well suited to be set to perform
5099 unattended refactoring. But if we could find a way to filter out the changes
5100 that create compilation errors, we could use the refactoring to look for
5101 improvement points in a software project. This process could for instance be
5102 scheduled to run at regular times, possibly after a nightly build or the like.
5103 Then the results could be made available, and an administrator could be set to
5104 review them and choose whether or not they should be performed.
5106 \section{Future work}
5107 An important part that is missing for making the search-based
5108 \ExtractAndMoveMethod refactoring more usable, is to complete the
5109 pre-refactoring analysis of the source code, to make sure that the refactoring
5110 does not introduce compilation errors when it is performed.
5112 The first point of making the static analysis complete brings up the next
5113 question: Is it feasible to complete such an analysis? Can this feasibility be
5114 proven, or disproved?
5116 Another shortcoming of this project is that we have no strategy for assuring
5117 safety when refactoring, so a program may end up behaving differently after a
5118 refactoring than it behaved before. One approach towards safer refactorings is
5119 mentioned in \myref{sec:saferRefactoringTests}, and includes generating tests
5120 for the refactored code. Another approach that can be considered for making
5121 refactorings safer is part of the original thesis proposal for this thesis,
5122 which diverged somewhat from the original proposal. The proposal is about
5123 detecting behavioral changes during refactoring, and the work done in this
5124 thesis can be used as a basis if one would like to engage in that proposal. The
5125 proposed solution for exposing behavioral changes is to insert assertions into
5126 source code when refactoring it. For the example in
5127 \myref{lst:correctnessExtractAndMoveResult}, which is the result of a
5128 refactoring, it is suggested that we insert an assert statement between lines 9
5129 and 10. In the example, the assert statement would be
5130 \mint{java}|assert c.x == this;| which would discover the aliasing problems of
5133 The final important improvement that I would suggest making to this project is
5134 to refine the heuristics that is used to find suitable refactoring candidates.
5135 This effort should in particular be directed toward making the heuristics choose
5136 candidates that do not introduce new dependencies for their destination classes.
5142 \chapter{Eclipse bugs submitted}
5143 \newcommand{\submittedBugReport}[1]{The submitted bug report can be found on
5146 \section{Eclipse bug 420726: Code is broken when moving a method that is
5147 assigning to the parameter that is also the move
5148 destination}\label{eclipse_bug_420726}
5150 was found when analyzing what kinds of names that were to be considered as
5151 \emph{unfixes} \see{unfixes}.
5154 The bug emerges when trying to move a method from one class to another, and when
5155 the target for the move (must be a variable, local or field) is both a parameter
5156 variable and also is assigned to within the method body. \name{Eclipse} allows this to
5157 happen, although it is the sure path to a compilation error. This is because we
5158 would then have an assignment to a \var{this} expression, which is not allowed
5160 \submittedBugReport{https://bugs.eclipse.org/bugs/show\_bug.cgi?id=420726}
5162 \paragraph{The solution}
5163 The solution to this problem is to add all simple names that are assigned to in
5164 a method body to the set of unfixes.
5166 \section{Eclipse bug 429416: IAE when moving method from anonymous
5167 class}\label{eclipse_bug_429416}
5169 this bug during a batch change on the \type{org.eclipse.jdt.ui} project.
5172 This bug surfaces when trying to use the \refa{Move Method} refactoring to move a
5173 method from an anonymous class to another class. This happens both for my
5174 simulation as well as in \name{Eclipse}, through the user interface. It only occurs
5175 when \name{Eclipse} analyzes the program and finds it necessary to pass an
5176 instance of the originating class as a parameter to the moved method. I.e. it
5177 wants to pass a \var{this} expression. The execution ends in an
5178 \typewithref{java.lang}{IllegalArgumentException} in
5179 \typewithref{org.eclipse.jdt.core.dom}{SimpleName} and its
5180 \method{setIdentifier(String)} method. The simple name is attempted created in
5182 \methodwithref{org.eclipse.jdt.internal.corext.refactoring.structure.\\MoveInstanceMethodProcessor}{createInlinedMethodInvocation}
5183 so the \type{MoveInstanceMethodProcessor} was early a clear suspect.
5185 The \method{createInlinedMethodInvocation} is the method that creates a method
5186 invocation where the previous invocation to the method that was moved was
5187 located. From its code it can be read that when a \var{this} expression is going
5188 to be passed in to the invocation, it shall be qualified with the name of the
5189 original method's declaring class, if the declaring class is either an anonymous
5190 class or a member class. The problem with this, is that an anonymous class does
5191 not have a name, hence the term \emph{anonymous} class! Therefore, when its
5192 name, an empty string, is passed into
5193 \methodwithref{org.eclipse.jdt.core.dom.AST}{newSimpleName} it all ends in an
5194 \type{IllegalArgumentException}.
5195 \submittedBugReport{https://bugs.eclipse.org/bugs/show\_bug.cgi?id=429416}
5197 \paragraph{How I solved the problem}
5198 Since the \type{MoveInstanceMethodProcessor} is instantiated in the
5199 \typewithref{no.uio.ifi.refaktor.change.executors}{MoveMethod\-RefactoringExecutor},
5200 and only need to be a
5201 \typewithref{org.eclipse.ltk.core.refactoring.participants}{MoveProcessor}, I
5202 was able to copy the code for the original move processor and modify it so that
5203 it works better for me. It is now called
5204 \typewithref{no.uio.ifi.refaktor.change.processors}{ModifiedMoveInstanceMethodProcessor}.
5205 The only modification done (in addition to some imports and suppression of
5206 warnings), is in the \method{createInlinedMethodInvocation}. When the declaring
5207 class of the method to move is anonymous, the \var{this} expression in the
5208 parameter list is not qualified with the declaring class' (empty) name.
5210 \section{Eclipse bug 429954: Extracting statement with reference to local type
5211 breaks code}\label{eclipse_bug_429954}
5212 The bug was discovered when doing some changes to the way unfixes is computed.
5215 The problem is that \name{Eclipse} is allowing selections that references variables of
5216 local types to be extracted. When this happens the code is broken, since the
5217 extracted method must take a parameter of a local type that is not in the
5218 methods scope. The problem is illustrated in
5219 \myref{lst:extractMethodLocalClass}, but there in another setting.
5220 \submittedBugReport{https://bugs.eclipse.org/bugs/show\_bug.cgi?id=429954}
5222 \paragraph{Actions taken}
5223 There are no actions directly springing out of this bug, since the Extract
5224 Method refactoring cannot be meant to be this way. This is handled on the
5225 analysis stage of our \refa{Extract and Move Method} refactoring. So names
5226 representing variables of local types are considered unfixes \see{unfixes}.