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
10 \usepackage{tikz-qtree}
11 \usetikzlibrary{shapes,snakes,trees,arrows,shadows,positioning,calc}
12 \usepackage{babel,textcomp,csquotes,ifimasterforside}
15 \usepackage[hidelinks]{hyperref}
18 \usepackage[style=alphabetic,backend=bibtex]{biblatex}
20 \usepackage{mathtools}
22 % use 'disable' before printing:
23 \usepackage[]{todonotes}
30 \usepackage{perpage} %the perpage package
31 \MakePerPage{footnote} %the perpage package command
33 \theoremstyle{definition}
34 \newtheorem*{wordDef}{Definition}
35 \newtheorem*{theorem}{Theorem}
37 \graphicspath{ {./figures/} }
39 \newcommand{\citing}[1]{~\cite{#1}}
40 %\newcommand{\myref}[1]{\cref{#1} on \cpageref{#1}}
41 \newcommand{\myref}[1]{\vref{#1}}
43 \newcommand{\definition}[1]{\begin{wordDef}#1\end{wordDef}}
44 \newcommand{\see}[1]{(see \myref{#1})}
45 \newcommand{\explanation}[3]{\noindent\textbf{\textit{#1}}\\*\emph{When:}
46 #2\\*\emph{How:} #3\\*[-7px]}
48 %\newcommand{\type}[1]{\lstinline{#1}}
49 \newcommand{\code}[1]{\texttt{\textbf{#1}}}
50 \newcommand{\type}[1]{\code{#1}}
51 \newcommand{\typeref}[1]{\footnote{\type{#1}}}
52 \newcommand{\typewithref}[2]{\type{#2}\typeref{#1.#2}}
53 \newcommand{\method}[1]{\type{#1}}
54 \newcommand{\methodref}[2]{\footnote{\type{#1}\method{\##2()}}}
55 \newcommand{\methodwithref}[2]{\method{#2}\footnote{\type{#1}\method{\##2()}}}
56 \newcommand{\var}[1]{\type{#1}}
58 \newcommand{\refactoring}[1]{\emph{#1}}
59 \newcommand{\ExtractMethod}{\refactoring{Extract Method}\xspace}
60 \newcommand{\MoveMethod}{\refactoring{Move Method}\xspace}
61 \newcommand{\ExtractAndMoveMethod}{\refactoring{Extract and Move Method}\xspace}
63 \newcommand\todoin[2][]{\todo[inline, caption={#2}, #1]{
64 \begin{minipage}{\textwidth-4pt}#2\end{minipage}}}
66 \title{Automated Composition of Refactorings}
67 \subtitle{Composing the Extract and Move Method refactorings in Eclipse}
68 \author{Erlend Kristiansen}
70 \bibliography{bibliography/master-thesis-erlenkr-bibliography}
72 % UML comment in TikZ:
73 % ref: https://tex.stackexchange.com/questions/103688/folded-paper-shape-tikz
75 \pgfdeclareshape{umlcomment}{
76 \inheritsavedanchors[from=rectangle] % this is nearly a rectangle
77 \inheritanchorborder[from=rectangle]
78 \inheritanchor[from=rectangle]{center}
79 \inheritanchor[from=rectangle]{north}
80 \inheritanchor[from=rectangle]{south}
81 \inheritanchor[from=rectangle]{west}
82 \inheritanchor[from=rectangle]{east}
83 % ... and possibly more
84 \backgroundpath{% this is new
85 % store lower right in xa/ya and upper right in xb/yb
86 \southwest \pgf@xa=\pgf@x \pgf@ya=\pgf@y
87 \northeast \pgf@xb=\pgf@x \pgf@yb=\pgf@y
88 % compute corner of ‘‘flipped page’’
89 \pgf@xc=\pgf@xb \advance\pgf@xc by-10pt % this should be a parameter
90 \pgf@yc=\pgf@yb \advance\pgf@yc by-10pt
92 \pgfpathmoveto{\pgfpoint{\pgf@xa}{\pgf@ya}}
93 \pgfpathlineto{\pgfpoint{\pgf@xa}{\pgf@yb}}
94 \pgfpathlineto{\pgfpoint{\pgf@xc}{\pgf@yb}}
95 \pgfpathlineto{\pgfpoint{\pgf@xb}{\pgf@yc}}
96 \pgfpathlineto{\pgfpoint{\pgf@xb}{\pgf@ya}}
99 \pgfpathmoveto{\pgfpoint{\pgf@xc}{\pgf@yb}}
100 \pgfpathlineto{\pgfpoint{\pgf@xc}{\pgf@yc}}
101 \pgfpathlineto{\pgfpoint{\pgf@xb}{\pgf@yc}}
102 \pgfpathlineto{\pgfpoint{\pgf@xc}{\pgf@yc}}
107 \tikzstyle{comment}=[%
126 \todoin{\textbf{Remove all todos (including list) before delivery/printing!!!
127 Can be done by removing ``draft'' from documentclass.}}
128 \todoin{Write abstract}
136 The discussions in this report must be seen in the context of object oriented
137 programming languages, and Java in particular, since that is the language in
138 which most of the examples will be given. All though the techniques discussed
139 may be applicable to languages from other paradigms, they will not be the
140 subject of this report.
144 \chapter{What is Refactoring?}
146 This question is best answered by first defining the concept of a
147 \emph{refactoring}, what it is to \emph{refactor}, and then discuss what aspects
148 of programming make people want to refactor their code.
150 \section{Defining refactoring}
151 Martin Fowler, in his classic book on refactoring\citing{refactoring}, defines a
152 refactoring like this:
155 \emph{Refactoring} (noun): a change made to the internal
156 structure\footnote{The structure observable by the programmer.} of software to
157 make it easier to understand and cheaper to modify without changing its
158 observable behavior.~\cite[p.~53]{refactoring}
161 \noindent This definition assigns additional meaning to the word
162 \emph{refactoring}, beyond the composition of the prefix \emph{re-}, usually
163 meaning something like ``again'' or ``anew'', and the word \emph{factoring},
164 that can mean to isolate the \emph{factors} of something. Here a \emph{factor}
165 would be close to the mathematical definition of something that divides a
166 quantity, without leaving a remainder. Fowler is mixing the \emph{motivation}
167 behind refactoring into his definition. Instead it could be more refined, formed
168 to only consider the \emph{mechanical} and \emph{behavioral} aspects of
169 refactoring. That is to factor the program again, putting it together in a
170 different way than before, while preserving the behavior of the program. An
171 alternative definition could then be:
173 \definition{A \emph{refactoring} is a transformation
174 done to a program without altering its external behavior.}
176 From this we can conclude that a refactoring primarily changes how the
177 \emph{code} of a program is perceived by the \emph{programmer}, and not the
178 \emph{behavior} experienced by any user of the program. Although the logical
179 meaning is preserved, such changes could potentially alter the program's
180 behavior when it comes to performance gain or -penalties. So any logic depending
181 on the performance of a program could make the program behave differently after
184 In the extreme case one could argue that such a thing as \emph{software
185 obfuscation} is refactoring. Software obfuscation is to make source code harder
186 to read and analyze, while preserving its semantics. It could be done composing
187 many, more or less randomly chosen, refactorings. Then the question arise
188 whether it can be called a \emph{composite refactoring}
189 \see{compositeRefactorings} or not? The answer is not obvious. First, there is
190 no way to describe \emph{the} mechanics of software obfuscation, beacause there
191 are infinitely many ways to do that. Second, \emph{obfuscation} can be thought
192 of as \emph{one operation}: Either the code is obfuscated, or it is not. Third,
193 it makes no sense to call software obfuscation \emph{a} refactoring, since it
194 holds different meaning to different people. The last point is important, since
195 one of the motivations behind defining different refactorings is to build up a
196 vocabulary for software professionals to reason and discuss about programs,
197 similar to the motivation behind design patterns\citing{designPatterns}. So for
198 describing \emph{software obfuscation}, it might be more appropriate to define
199 what you do when performing it rather than precisely defining its mechanics in
200 terms of other refactorings.
202 \section{The etymology of 'refactoring'}
203 It is a little difficult to pinpoint the exact origin of the word
204 ``refactoring'', as it seems to have evolved as part of a colloquial
205 terminology, more than a scientific term. There is no authoritative source for a
206 formal definition of it.
208 According to Martin Fowler\citing{etymology-refactoring}, there may also be more
209 than one origin of the word. The most well-known source, when it comes to the
210 origin of \emph{refactoring}, is the Smalltalk\footnote{\emph{Smalltalk},
211 object-oriented, dynamically typed, reflective programming language. See
212 \url{http://www.smalltalk.org}} community and their infamous \emph{Refactoring
213 Browser}\footnote{\url{http://st-www.cs.illinois.edu/users/brant/Refactory/RefactoringBrowser.html}}
214 described in the article \emph{A Refactoring Tool for
215 Smalltalk}\citing{refactoringBrowser1997}, published in 1997.
216 Allegedly\citing{etymology-refactoring}, the metaphor of factoring programs was
217 also present in the Forth\footnote{\emph{Forth} -- stack-based, extensible
218 programming language, without type-checking. See \url{http://www.forth.org}}
219 community, and the word ``refactoring'' is mentioned in a book by Leo Brodie,
220 called \emph{Thinking Forth}\citing{brodie2004}, first published in
221 1984\footnote{\emph{Thinking Forth} was first published in 1984 by the
222 \emph{Forth Interest Group}. Then it was reprinted in 1994 with minor
223 typographical corrections, before it was transcribed into an electronic edition
224 typeset in \LaTeX\ and published under a Creative Commons licence in 2004. The
225 edition cited here is the 2004 edition, but the content should essentially be as
226 in 1984.}. The exact word is only printed one place~\cite[p.~232]{brodie2004},
227 but the term \emph{factoring} is prominent in the book, that also contains a
228 whole chapter dedicated to (re)factoring, and how to keep the (Forth) code clean
232 \ldots good factoring technique is perhaps the most important skill for a
233 Forth programmer.~\cite[p.~172]{brodie2004}
236 \noindent Brodie also express what \emph{factoring} means to him:
239 Factoring means organizing code into useful fragments. To make a fragment
240 useful, you often must separate reusable parts from non-reusable parts. The
241 reusable parts become new definitions. The non-reusable parts become arguments
242 or parameters to the definitions.~\cite[p.~172]{brodie2004}
245 Fowler claims that the usage of the word \emph{refactoring} did not pass between
246 the \emph{Forth} and \emph{Smalltalk} communities, but that it emerged
247 independently in each of the communities.
249 \section{Motivation -- Why people refactor}
250 There are many reasons why people want to refactor their programs. They can for
251 instance do it to remove duplication, break up long methods or to introduce
252 design patterns\citing{designPatterns} into their software systems. The shared
253 trait for all these are that peoples intentions are to make their programs
254 \emph{better}, in some sense. But what aspects of their programs are becoming
257 As already mentioned, people often refactor to get rid of duplication. Moving
258 identical or similar code into methods, and maybe pushing methods up or down in
259 their class hierarchies. Making template methods for overlapping
260 algorithms/functionality and so on. It is all about gathering what belongs
261 together and putting it all in one place. The resulting code is then easier to
262 maintain. When removing the implicit coupling\footnote{When duplicating code,
263 the code might not be coupled in other ways than that it is supposed to
264 represent the same functionality. So if this functionality is going to change,
265 it might need to change in more than one place, thus creating an implicit
266 coupling between the multiple pieces of code.} between code snippets, the
267 location of a bug is limited to only one place, and new functionality need only
268 to be added to this one place, instead of a number of places people might not
271 A problem you often encounter when programming, is that a program contains a lot
272 of long and hard-to-grasp methods. It can then help to break the methods into
273 smaller ones, using the \ExtractMethod refactoring\citing{refactoring}. Then you
274 may discover something about a program that you were not aware of before;
275 revealing bugs you did not know about or could not find due to the complex
276 structure of your program. \todo{Proof?} Making the methods smaller and giving
277 good names to the new ones clarifies the algorithms and enhances the
278 \emph{understandability} of the program \see{magic_number_seven}. This makes
279 refactoring an excellent method for exploring unknown program code, or code that
280 you had forgotten that you wrote.
282 Most primitive refactorings are simple. Their true power is first revealed when
283 they are combined into larger --- higher level --- refactorings, called
284 \emph{composite refactorings} \see{compositeRefactorings}. Often the goal of
285 such a series of refactorings is a design pattern. Thus the \emph{design} can be
286 evolved throughout the lifetime of a program, as opposed to designing up-front.
287 It is all about being structured and taking small steps to improve a program's
290 Many software design pattern are aimed at lowering the coupling between
291 different classes and different layers of logic. One of the most famous is
292 perhaps the \emph{Model-View-Controller}\citing{designPatterns} pattern. It is
293 aimed at lowering the coupling between the user interface and the business logic
294 and data representation of a program. This also has the added benefit that the
295 business logic could much easier be the target of automated tests, increasing
296 the productivity in the software development process. Refactoring is an
297 important tool on the way to something greater.
299 Another effect of refactoring is that with the increased separation of concerns
300 coming out of many refactorings, the \emph{performance} can be improved. When
301 profiling programs, the problematic parts are narrowed down to smaller parts of
302 the code, which are easier to tune, and optimization can be performed only where
303 needed and in a more effective way.
305 Last, but not least, and this should probably be the best reason to refactor, is
306 to refactor to \emph{facilitate a program change}. If one has managed to keep
307 one's code clean and tidy, and the code is not bloated with design patterns that
308 are not ever going to be needed, then some refactoring might be needed to
309 introduce a design pattern that is appropriate for the change that is going to
312 Refactoring program code --- with a goal in mind --- can give the code itself
313 more value. That is in the form of robustness to bugs, understandability and
314 maintainability. Having robust code is an obvious advantage, but
315 understandability and maintainability are both very important aspects of
316 software development. By incorporating refactoring in the development process,
317 bugs are found faster, new functionality is added more easily and code is easier
318 to understand by the next person exposed to it, which might as well be the
319 person who wrote it. The consequence of this, is that refactoring can increase
320 the average productivity of the development process, and thus also add to the
321 monetary value of a business in the long run. The perspective on productivity
322 and money should also be able to open the eyes of the many nearsighted managers
323 that seldom see beyond the next milestone.
325 \section{The magical number seven}\label{magic_number_seven}
326 The article \emph{The magical number seven, plus or minus two: some limits on
327 our capacity for processing information}\citing{miller1956} by George A.
328 Miller, was published in the journal \emph{Psychological Review} in 1956. It
329 presents evidence that support that the capacity of the number of objects a
330 human being can hold in its working memory is roughly seven, plus or minus two
331 objects. This number varies a bit depending on the nature and complexity of the
332 objects, but is according to Miller ``\ldots never changing so much as to be
335 Miller's article culminates in the section called \emph{Recoding}, a term he
336 borrows from communication theory. The central result in this section is that by
337 recoding information, the capacity of the amount of information that a human can
338 process at a time is increased. By \emph{recoding}, Miller means to group
339 objects together in chunks and give each chunk a new name that it can be
340 remembered by. By organizing objects into patterns of ever growing depth, one
341 can memorize and process a much larger amount of data than if it were to be
342 represented as its basic pieces. This grouping and renaming is analogous to how
343 many refactorings work, by grouping pieces of code and give them a new name.
344 Examples are the fundamental \ExtractMethod and \refactoring{Extract Class}
345 refactorings\citing{refactoring}.
348 \ldots recoding is an extremely powerful weapon for increasing the amount of
349 information that we can deal with.~\cite[p.~95]{miller1956}
352 An example from the article addresses the problem of memorizing a sequence of
353 binary digits. Let us say we have the following sequence\footnote{The example
354 presented here is slightly modified (and shortened) from what is presented in
355 the original article\citing{miller1956}, but it is essentially the same.} of
356 16 binary digits: ``1010001001110011''. Most of us will have a hard time
357 memorizing this sequence by only reading it once or twice. Imagine if we instead
358 translate it to this sequence: ``A273''. If you have a background from computer
359 science, it will be obvious that the latest sequence is the first sequence
360 recoded to be represented by digits with base 16. Most people should be able to
361 memorize this last sequence by only looking at it once.
363 Another result from the Miller article is that when the amount of information a
364 human must interpret increases, it is crucial that the translation from one code
365 to another must be almost automatic for the subject to be able to remember the
366 translation, before \heshe is presented with new information to recode. Thus
367 learning and understanding how to best organize certain kinds of data is
368 essential to efficiently handle that kind of data in the future. This is much
369 like when humans learn to read. First they must learn how to recognize letters.
370 Then they can learn distinct words, and later read sequences of words that form
371 whole sentences. Eventually, most of them will be able to read whole books and
372 briefly retell the important parts of its content. This suggest that the use of
373 design patterns\citing{designPatterns} is a good idea when reasoning about
374 computer programs. With extensive use of design patterns when creating complex
375 program structures, one does not always have to read whole classes of code to
376 comprehend how they function, it may be sufficient to only see the name of a
377 class to almost fully understand its responsibilities.
380 Our language is tremendously useful for repackaging material into a few chunks
381 rich in information.~\cite[p.~95]{miller1956}
384 Without further evidence, these results at least indicate that refactoring
385 source code into smaller units with higher cohesion and, when needed,
386 introducing appropriate design patterns, should aid in the cause of creating
387 computer programs that are easier to maintain and has code that is easier (and
390 \section{Notable contributions to the refactoring literature}
391 \todoin{Update with more contributions}
394 \item[1992] William F. Opdyke submits his doctoral dissertation called
395 \emph{Refactoring Object-Oriented Frameworks}\citing{opdyke1992}. This
396 work defines a set of refactorings, that are behavior preserving given that
397 their preconditions are met. The dissertation is focused on the automation
399 \item[1999] Martin Fowler et al.: \emph{Refactoring: Improving the Design of
400 Existing Code}\citing{refactoring}. This is maybe the most influential text
401 on refactoring. It bares similarities with Opdykes thesis\citing{opdyke1992}
402 in the way that it provides a catalog of refactorings. But Fowler's book is
403 more about the craft of refactoring, as he focuses on establishing a
404 vocabulary for refactoring, together with the mechanics of different
405 refactorings and when to perform them. His methodology is also founded on
406 the principles of test-driven development.
407 \item[2005] Joshua Kerievsky: \emph{Refactoring to
408 Patterns}\citing{kerievsky2005}. This book is heavily influenced by Fowler's
409 \emph{Refactoring}\citing{refactoring} and the ``Gang of Four'' \emph{Design
410 Patterns}\citing{designPatterns}. It is building on the refactoring
411 catalogue from Fowler's book, but is trying to bridge the gap between
412 \emph{refactoring} and \emph{design patterns} by providing a series of
413 higher-level composite refactorings, that makes code evolve toward or away
414 from certain design patterns. The book is trying to build up the readers
415 intuition around \emph{why} one would want to use a particular design
416 pattern, and not just \emph{how}. The book is encouraging evolutionary
417 design \see{relationToDesignPatterns}.
420 \section{Tool support (for Java)}\label{toolSupport}
421 This section will briefly compare the refatoring support of the three IDEs
422 \emph{Eclipse}\footnote{\url{http://www.eclipse.org/}}, \emph{IntelliJ
423 IDEA}\footnote{The IDE under comparison is the \emph{Community Edition},
424 \url{http://www.jetbrains.com/idea/}} and
425 \emph{NetBeans}\footnote{\url{https://netbeans.org/}}. These are the most
426 popular Java IDEs\citing{javaReport2011}.
428 All three IDEs provide support for the most useful refactorings, like the
429 different extract, move and rename refactorings. In fact, Java-targeted IDEs are
430 known for their good refactoring support, so this did not appear as a big
433 The IDEs seem to have excellent support for the \ExtractMethod refactoring, so
434 at least they have all passed the first refactoring
435 rubicon\citing{fowlerRubicon2001,secondRubicon2012}.
437 Regarding the \MoveMethod refactoring, the \emph{Eclipse} and \emph{IntelliJ}
438 IDEs do the job in very similar manners. In most situations they both do a
439 satisfying job by producing the expected outcome. But they do nothing to check
440 that the result does not break the semantics of the program \see{correctness}.
441 The \emph{NetBeans} IDE implements this refactoring in a somewhat
442 unsophisticated way. For starters, its default destination for the move is
443 itself, although it refuses to perform the refactoring if chosen. But the worst
444 part is, that if moving the method \method{f} of the class \type{C} to the class
445 \type{X}, it will break the code. The result is shown in
446 \myref{lst:moveMethod_NetBeans}.
450 \begin{minted}[samepage]{java}
463 \begin{minted}[samepage]{java}
473 \caption{Moving method \method{f} from \type{C} to \type{X}.}
474 \label{lst:moveMethod_NetBeans}
477 NetBeans will try to make code that call the methods \method{m} and \method{n}
478 of \type{X} by accessing them through \var{c.x}, where \var{c} is a parameter of
479 type \type{C} that is added the method \method{f} when it is moved. (This is
480 seldom the desired outcome of this refactoring, but ironically, this ``feature''
481 keeps NetBeans from breaking the code in the example from \myref{correctness}.)
482 If \var{c.x} for some reason is inaccessible to \type{X}, as in this case, the
483 refactoring breaks the code, and it will not compile. NetBeans presents a
484 preview of the refactoring outcome, but the preview does not catch it if the IDE
485 is about break the program.
487 The IDEs under investigation seems to have fairly good support for primitive
488 refactorings, but what about more complex ones, such as the \refactoring{Extract
489 Class}\citing{refactoring}? The \refactoring{Extract Class} refactoring works by
490 creating a class, for then to move members to that class and access them from
491 the old class via a reference to the new class. \emph{IntelliJ} handles this in
492 a fairly good manner, although, in the case of private methods, it leaves unused
493 methods behind. These are methods that delegate to a field with the type of the
494 new class, but are not used anywhere. \emph{Eclipse} has added (or withdrawn)
495 its own quirk to the Extract Class refactoring, and only allows for
496 \emph{fields} to be moved to a new class, \emph{not methods}. This makes it
497 effectively only extracting a data structure, and calling it
498 \refactoring{Extract Class} is a little misleading. One would often be better
499 off with textual extract and paste than using the Extract Class refactoring in
500 Eclipse. When it comes to \emph{NetBeans}, it does not even seem to have made an
501 attempt on providing this refactoring. (Well, it probably has, but it does not
504 \todoin{Visual Studio (C++/C\#), Smalltalk refactoring browser?,
505 second refactoring rubicon?}
507 \section{The relation to design patterns}\label{relationToDesignPatterns}
509 \emph{Refactoring} and \emph{design patterns} have at least one thing in common,
510 they are both promoted by advocates of \emph{clean code}\citing{cleanCode} as
511 fundamental tools on the road to more maintanable and extendable source code.
514 Design patterns help you determine how to reorganize a design, and they can
515 reduce the amount of refactoring you need to do
516 later.~\cite[p.~353]{designPatterns}
519 Although sometimes associated with
520 over-engineering\citing{kerievsky2005,refactoring}, design patterns are in
521 general assumed to be good for maintainability of source code. That may be
522 because many of them are designed to support the \emph{open/closed principle} of
523 object-oriented programming. The principle was first formulated by Bertrand
524 Meyer, the creator of the Eiffel programming language, like this: ``Modules
525 should be both open and closed.''\citing{meyer1988} It has been popularized,
526 with this as a common version:
529 Software entities (classes, modules, functions, etc.) should be open for
530 extension, but closed for modification.\footnote{See
531 \url{http://c2.com/cgi/wiki?OpenClosedPrinciple} or
532 \url{https://en.wikipedia.org/wiki/Open/closed_principle}}
535 Maintainability is often thought of as the ability to be able to introduce new
536 functionality without having to change too much of the old code. When
537 refactoring, the motivation is often to facilitate adding new functionality. It
538 is about factoring the old code in a way that makes the new functionality being
539 able to benefit from the functionality already residing in a software system,
540 without having to copy old code into new. Then, next time someone shall add new
541 functionality, it is less likely that the old code has to change. Assuming that
542 a design pattern is the best way to get rid of duplication and assist in
543 implementing new functionality, it is reasonable to conclude that a design
544 pattern often is the target of a series of refactorings. Having a repertoire of
545 design patterns can also help in knowing when and how to refactor a program to
546 make it reflect certain desired characteristics.
549 There is a natural relation between patterns and refactorings. Patterns are
550 where you want to be; refactorings are ways to get there from somewhere
551 else.~\cite[p.~107]{refactoring}
554 This quote is wise in many contexts, but it is not always appropriate to say
555 ``Patterns are where you want to be\ldots''. \emph{Sometimes}, patterns are
556 where you want to be, but only because it will benefit your design. It is not
557 true that one should always try to incorporate as many design patterns as
558 possible into a program. It is not like they have intrinsic value. They only add
559 value to a system when they support its design. Otherwise, the use of design
560 patterns may only lead to a program that is more complex than necessary.
563 The overuse of patterns tends to result from being patterns happy. We are
564 \emph{patterns happy} when we become so enamored of patterns that we simply
565 must use them in our code.~\cite[p.~24]{kerievsky2005}
568 This can easily happen when relying largely on up-front design. Then it is
569 natural, in the very beginning, to try to build in all the flexibility that one
570 believes will be necessary throughout the lifetime of a software system.
571 According to Joshua Kerievsky ``That sounds reasonable --- if you happen to be
572 psychic.''~\cite[p.~1]{kerievsky2005} He is advocating what he believes is a
573 better approach: To let software continually evolve. To start with a simple
574 design that meets today's needs, and tackle future needs by refactoring to
575 satisfy them. He believes that this is a more economic approach than investing
576 time and money into a design that inevitably is going to change. By relying on
577 continuously refactoring a system, its design can be made simpler without
578 sacrificing flexibility. To be able to fully rely on this approach, it is of
579 utter importance to have a reliable suit of tests to lean on \see{testing}. This
580 makes the design process more natural and less characterized by difficult
581 decisions that has to be made before proceeding in the process, and that is
582 going to define a project for all of its unforeseeable future.
586 \section{Classification of refactorings}
587 % only interesting refactorings
588 % with 2 detailed examples? One for structured and one for intra-method?
589 % Is replacing Bubblesort with Quick Sort considered a refactoring?
591 \subsection{Structural refactorings}
593 \subsubsection{Primitive refactorings}
596 \explanation{Extract Method}{You have a code fragment that can be grouped
597 together.}{Turn the fragment into a method whose name explains the purpose of
600 \explanation{Inline Method}{A method's body is just as clear as its name.}{Put
601 the method's body into the body of its callers and remove the method.}
603 \explanation{Inline Temp}{You have a temp that is assigned to once with a simple
604 expression, and the temp is getting in the way of other refactorings.}{Replace
605 all references to that temp with the expression}
607 % Moving Features Between Objects
608 \explanation{Move Method}{A method is, or will be, using or used by more
609 features of another class than the class on which it is defined.}{Create a new
610 method with a similar body in the class it uses most. Either turn the old method
611 into a simple delegation, or remove it altogether.}
613 \explanation{Move Field}{A field is, or will be, used by another class more than
614 the class on which it is defined}{Create a new field in the target class, and
615 change all its users.}
618 \explanation{Replace Magic Number with Symbolic Constant}{You have a literal
619 number with a particular meaning.}{Create a constant, name it after the meaning,
620 and replace the number with it.}
622 \explanation{Encapsulate Field}{There is a public field.}{Make it private and
625 \explanation{Replace Type Code with Class}{A class has a numeric type code that
626 does not affect its behavior.}{Replace the number with a new class.}
628 \explanation{Replace Type Code with Subclasses}{You have an immutable type code
629 that affects the behavior of a class.}{Replace the type code with subclasses.}
631 \explanation{Replace Type Code with State/Strategy}{You have a type code that
632 affects the behavior of a class, but you cannot use subclassing.}{Replace the
633 type code with a state object.}
635 % Simplifying Conditional Expressions
636 \explanation{Consolidate Duplicate Conditional Fragments}{The same fragment of
637 code is in all branches of a conditional expression.}{Move it outside of the
640 \explanation{Remove Control Flag}{You have a variable that is acting as a
641 control flag fro a series of boolean expressions.}{Use a break or return
644 \explanation{Replace Nested Conditional with Guard Clauses}{A method has
645 conditional behavior that does not make clear the normal path of
646 execution.}{Use guard clauses for all special cases.}
648 \explanation{Introduce Null Object}{You have repeated checks for a null
649 value.}{Replace the null value with a null object.}
651 \explanation{Introduce Assertion}{A section of code assumes something about the
652 state of the program.}{Make the assumption explicit with an assertion.}
654 % Making Method Calls Simpler
655 \explanation{Rename Method}{The name of a method does not reveal its
656 purpose.}{Change the name of the method}
658 \explanation{Add Parameter}{A method needs more information from its
659 caller.}{Add a parameter for an object that can pass on this information.}
661 \explanation{Remove Parameter}{A parameter is no longer used by the method
664 %\explanation{Parameterize Method}{Several methods do similar things but with
665 %different values contained in the method.}{Create one method that uses a
666 %parameter for the different values.}
668 \explanation{Preserve Whole Object}{You are getting several values from an
669 object and passing these values as parameters in a method call.}{Send the whole
672 \explanation{Remove Setting Method}{A field should be set at creation time and
673 never altered.}{Remove any setting method for that field.}
675 \explanation{Hide Method}{A method is not used by any other class.}{Make the
678 \explanation{Replace Constructor with Factory Method}{You want to do more than
679 simple construction when you create an object}{Replace the constructor with a
682 % Dealing with Generalization
683 \explanation{Pull Up Field}{Two subclasses have the same field.}{Move the field
686 \explanation{Pull Up Method}{You have methods with identical results on
687 subclasses.}{Move them to the superclass.}
689 \explanation{Push Down Method}{Behavior on a superclass is relevant only for
690 some of its subclasses.}{Move it to those subclasses.}
692 \explanation{Push Down Field}{A field is used only by some subclasses.}{Move the
693 field to those subclasses}
695 \explanation{Extract Interface}{Several clients use the same subset of a class's
696 interface, or two classes have part of their interfaces in common.}{Extract the
697 subset into an interface.}
699 \explanation{Replace Inheritance with Delegation}{A subclass uses only part of a
700 superclasses interface or does not want to inherit data.}{Create a field for the
701 superclass, adjust methods to delegate to the superclass, and remove the
704 \explanation{Replace Delegation with Inheritance}{You're using delegation and
705 are often writing many simple delegations for the entire interface}{Make the
706 delegating class a subclass of the delegate.}
708 \subsubsection{Composite refactorings}
711 % \explanation{Replace Method with Method Object}{}{}
713 % Moving Features Between Objects
714 \explanation{Extract Class}{You have one class doing work that should be done by
715 two}{Create a new class and move the relevant fields and methods from the old
716 class into the new class.}
718 \explanation{Inline Class}{A class isn't doing very much.}{Move all its features
719 into another class and delete it.}
721 \explanation{Hide Delegate}{A client is calling a delegate class of an
722 object.}{Create Methods on the server to hide the delegate.}
724 \explanation{Remove Middle Man}{A class is doing to much simple delegation.}{Get
725 the client to call the delegate directly.}
728 \explanation{Replace Data Value with Object}{You have a data item that needs
729 additional data or behavior.}{Turn the data item into an object.}
731 \explanation{Change Value to Reference}{You have a class with many equal
732 instances that you want to replace with a single object.}{Turn the object into a
735 \explanation{Encapsulate Collection}{A method returns a collection}{Make it
736 return a read-only view and provide add/remove methods.}
738 % \explanation{Replace Array with Object}{}{}
740 \explanation{Replace Subclass with Fields}{You have subclasses that vary only in
741 methods that return constant data.}{Change the methods to superclass fields and
742 eliminate the subclasses.}
744 % Simplifying Conditional Expressions
745 \explanation{Decompose Conditional}{You have a complicated conditional
746 (if-then-else) statement.}{Extract methods from the condition, then part, an
749 \explanation{Consolidate Conditional Expression}{You have a sequence of
750 conditional tests with the same result.}{Combine them into a single conditional
751 expression and extract it.}
753 \explanation{Replace Conditional with Polymorphism}{You have a conditional that
754 chooses different behavior depending on the type of an object.}{Move each leg
755 of the conditional to an overriding method in a subclass. Make the original
758 % Making Method Calls Simpler
759 \explanation{Replace Parameter with Method}{An object invokes a method, then
760 passes the result as a parameter for a method. The receiver can also invoke this
761 method.}{Remove the parameter and let the receiver invoke the method.}
763 \explanation{Introduce Parameter Object}{You have a group of parameters that
764 naturally go together.}{Replace them with an object.}
766 % Dealing with Generalization
767 \explanation{Extract Subclass}{A class has features that are used only in some
768 instances.}{Create a subclass for that subset of features.}
770 \explanation{Extract Superclass}{You have two classes with similar
771 features.}{Create a superclass and move the common features to the
774 \explanation{Collapse Hierarchy}{A superclass and subclass are not very
775 different.}{Merge them together.}
777 \explanation{Form Template Method}{You have two methods in subclasses that
778 perform similar steps in the same order, yet the steps are different.}{Get the
779 steps into methods with the same signature, so that the original methods become
780 the same. Then you can pull them up.}
783 \subsection{Functional refactorings}
785 \explanation{Substitute Algorithm}{You want to replace an algorithm with one
786 that is clearer.}{Replace the body of the method with the new algorithm.}
790 \section{The impact on software quality}
792 \subsection{What is software quality?}
793 The term \emph{software quality} has many meanings. It all depends on the
794 context we put it in. If we look at it with the eyes of a software developer, it
795 usually means that the software is easily maintainable and testable, or in other
796 words, that it is \emph{well designed}. This often correlates with the
797 management scale, where \emph{keeping the schedule} and \emph{customer
798 satisfaction} is at the center. From the customers point of view, in addition to
799 good usability, \emph{performance} and \emph{lack of bugs} is always
800 appreciated, measurements that are also shared by the software developer. (In
801 addition, such things as good documentation could be measured, but this is out
802 of the scope of this document.)
804 \subsection{The impact on performance}
806 Refactoring certainly will make software go more slowly\footnote{With todays
807 compiler optimization techniques and performance tuning of e.g. the Java
808 virtual machine, the penalties of object creation and method calls are
809 debatable.}, but it also makes the software more amenable to performance
810 tuning.~\cite[p.~69]{refactoring}
813 \noindent There is a common belief that refactoring compromises performance, due
814 to increased degree of indirection and that polymorphism is slower than
817 In a survey, Demeyer\citing{demeyer2002} disproves this view in the case of
818 polymorphism. He did an experiment on, what he calls, ``Transform Self Type
819 Checks'' where you introduce a new polymorphic method and a new class hierarchy
820 to get rid of a class' type checking of a ``type attribute``. He uses this kind
821 of transformation to represent other ways of replacing conditionals with
822 polymorphism as well. The experiment is performed on the C++ programming
823 language and with three different compilers and platforms. Demeyer concludes
824 that, with compiler optimization turned on, polymorphism beats middle to large
825 sized if-statements and does as well as case-statements. (In accordance with
826 his hypothesis, due to similarities between the way C++ handles polymorphism and
830 The interesting thing about performance is that if you analyze most programs,
831 you find that they waste most of their time in a small fraction of the
832 code.~\cite[p.~70]{refactoring}
835 \noindent So, although an increased amount of method calls could potentially
836 slow down programs, one should avoid premature optimization and sacrificing good
837 design, leaving the performance tuning until after profiling\footnote{For and
838 example of a Java profiler, check out VisualVM:
839 \url{http://visualvm.java.net/}} the software and having isolated the actual
842 \section{Composite refactorings}\label{compositeRefactorings}
843 \todo{motivation, examples, manual vs automated?, what about refactoring in a
844 very large code base?}
845 Generally, when thinking about refactoring, at the mechanical level, there are
846 essentially two kinds of refactorings. There are the \emph{primitive}
847 refactorings, and the \emph{composite} refactorings.
849 \definition{A \emph{primitive refactoring} is a refactoring that cannot be
850 expressed in terms of other refactorings.}
852 \noindent Examples are the \refactoring{Pull Up Field} and \refactoring{Pull Up
853 Method} refactorings\citing{refactoring}, that move members up in their class
856 \definition{A \emph{composite refactoring} is a refactoring that can be
857 expressed in terms of two or more other refactorings.}
859 \noindent An example of a composite refactoring is the \refactoring{Extract
860 Superclass} refactoring\citing{refactoring}. In its simplest form, it is composed
861 of the previously described primitive refactorings, in addition to the
862 \refactoring{Pull Up Constructor Body} refactoring\citing{refactoring}. It works
863 by creating an abstract superclass that the target class(es) inherits from, then
864 by applying \refactoring{Pull Up Field}, \refactoring{Pull Up Method} and
865 \refactoring{Pull Up Constructor Body} on the members that are to be members of
866 the new superclass. For an overview of the \refactoring{Extract Superclass}
867 refactoring, see \myref{fig:extractSuperclass}.
871 \includegraphics[angle=270,width=\linewidth]{extractSuperclassItalic.pdf}
872 \caption{The Extract Superclass refactoring}
873 \label{fig:extractSuperclass}
876 \section{Manual vs. automated refactorings}
877 Refactoring is something every programmer does, even if \heshe does not known
878 the term \emph{refactoring}. Every refinement of source code that does not alter
879 the program's behavior is a refactoring. For small refactorings, such as
880 \ExtractMethod, executing it manually is a manageable task, but is still prone
881 to errors. Getting it right the first time is not easy, considering the method
882 signature and all the other aspects of the refactoring that has to be in place.
884 Take for instance the renaming of classes, methods and fields. For complex
885 programs these refactorings are almost impossible to get right. Attacking them
886 with textual search and replace, or even regular expressions, will fall short on
887 these tasks. Then it is crucial to have proper tool support that can perform
888 them automatically. Tools that can parse source code and thus have semantic
889 knowledge about which occurrences of which names belong to what construct in the
890 program. For even trying to perform one of these complex task manually, one
891 would have to be very confident on the existing test suite \see{testing}.
893 \section{Correctness of refactorings}\label{correctness}
894 For automated refactorings to be truly useful, they must show a high degree of
895 behavior preservation. This last sentence might seem obvious, but there are
896 examples of refactorings in existing tools that break programs. I will now
897 present an example of an \ExtractMethod refactoring followed by a \MoveMethod
898 refactoring that breaks a program in both the \emph{Eclipse} and \emph{IntelliJ}
899 IDEs\footnote{The NetBeans IDE handles this particular situation without
900 altering the program's beavior, mainly because its Move Method refactoring
901 implementation is a bit flawed in other ways \see{toolSupport}.}. The
902 following piece of code shows the target for the composed refactoring:
904 \begin{minted}[linenos,samepage]{java}
906 public X x = new X();
915 \noindent The next piece of code shows the destination of the refactoring. Note
916 that the method \method{m(C c)} of class \type{C} assigns to the field \var{x}
917 of the argument \var{c} that has type \type{C}:
919 \begin{minted}[samepage]{java}
928 The refactoring sequence works by extracting line 5 and 6 from the original
929 class \type{C} into a method \method{f} with the statements from those lines as
930 its method body. The method is then moved to the class \type{X}. The result is
931 shown in the following two pieces of code:
933 \begin{minted}[linenos,samepage]{java}
935 public X x = new X();
943 \begin{minted}[linenos,samepage]{java}
956 After the refactoring, the method \method{f} of class \type{C} is calling the
957 method \method{f} of class \type{X}, and the program now behaves different than
958 before. (See line 5 of the version of class \type{C} after the refactoring.)
959 Before the refactoring, the methods \method{m} and \method{n} of class \type{X}
960 are called on different object instances (see line 5 and 6 of the original class
961 \type{C}). After, they are called on the same object, and the statement on line
962 3 of class \type{X} (the version after the refactoring) no longer have any
963 effect in our example.
965 The bug introduced in the previous example is of such a nature\footnote{Caused
966 by aliasing. See \url{https://en.wikipedia.org/wiki/Aliasing_(computing)}}
967 that it is very difficult to spot if the refactored code is not covered by
968 tests. It does not generate compilation errors, and will thus only result in
969 a runtime error or corrupted data, which might be hard to detect.
971 \section{Refactoring and the importance of testing}\label{testing}
973 If you want to refactor, the essential precondition is having solid
974 tests.\citing{refactoring}
977 When refactoring, there are roughly three classes of errors that can be made.
978 The first class of errors are the ones that make the code unable to compile.
979 These \emph{compile-time} errors are of the nicer kind. They flash up at the
980 moment they are made (at least when using an IDE), and are usually easy to fix.
981 The second class are the \emph{runtime} errors. Although they take a bit longer
982 to surface, they usually manifest after some time in an illegal argument
983 exception, null pointer exception or similar during the program execution.
984 These kind of errors are a bit harder to handle, but at least they will show,
985 eventually. Then there are the \emph{behavior-changing} errors. These errors are
986 of the worst kind. They do not show up during compilation and they do not turn
987 on a blinking red light during runtime either. The program can seem to work
988 perfectly fine with them in play, but the business logic can be damaged in ways
989 that will only show up over time.
991 For discovering runtime errors and behavior changes when refactoring, it is
992 essential to have good test coverage. Testing in this context means writing
993 automated tests. Manual testing may have its uses, but when refactoring, it is
994 automated unit testing that dominate. For discovering behavior changes it is
995 especially important to have tests that cover potential problems, since these
996 kind of errors does not reveal themselves.
998 Unit testing is not a way to \emph{prove} that a program is correct, but it is a
999 way to make you confindent that it \emph{probably} works as desired. In the
1000 context of test driven development (commonly known as TDD), the tests are even a
1001 way to define how the program is \emph{supposed} to work. It is then, by
1002 definition, working if the tests are passing.
1004 If the test coverage for a code base is perfect, then it should, theoretically,
1005 be risk-free to perform refactorings on it. This is why automated tests and
1006 refactoring are such a great match.
1008 \subsection{Testing the code from correctness section}
1009 The worst thing that can happen when refactoring is to introduce changes to the
1010 behavior of a program, as in the example on \myref{correctness}. This example
1011 may be trivial, but the essence is clear. The only problem with the example is
1012 that it is not clear how to create automated tests for it, without changing it
1015 Unit tests, as they are known from the different xUnit frameworks around, are
1016 only suitable to test the \emph{result} of isolated operations. They can not
1017 easily (if at all) observe the \emph{history} of a program.
1019 This problem is still open.
1024 Assuming a sequential (non-concurrent) program:
1026 \begin{minted}{java}
1027 tracematch (C c, X x) {
1029 call(* X.m(C)) && args(c) && cflow(within(C));
1031 call(* X.n()) && target(x) && cflow(within(C));
1033 set(C.x) && target(c) && !cflow(m);
1037 { assert x == c.x; }
1041 %\begin{minted}{java}
1042 %tracematch (X x1, X x2) {
1044 % call(* X.m(C)) && target(x1);
1046 % call(* X.n()) && target(x2);
1048 % set(C.x) && !cflow(m) && !cflow(n);
1052 % { assert x1 != x2; }
1057 \section{The project}
1058 The aim of this master project will be to investigate the relationship between a
1059 composite refactoring composed of the \ExtractMethod and \MoveMethod
1060 refactorings, and its impact on one or more software metrics.
1062 The composition of the \ExtractMethod and \MoveMethod refactorings springs
1063 naturally out of the need to move procedures closer to the data they manipulate.
1064 This composed refactoring is not well described in the literature, but it is
1065 implemented in at least one tool called
1066 \emph{CodeRush}\footnote{\url{https://help.devexpress.com/\#CodeRush/CustomDocument3519}},
1067 that is an extension for \emph{MS Visual
1068 Studio}\footnote{\url{http://www.visualstudio.com/}}. In CodeRush it is called
1069 \emph{Extract Method to
1070 Type}\footnote{\url{https://help.devexpress.com/\#CodeRush/CustomDocument6710}},
1071 but I choose to call it \ExtractAndMoveMethod, since I feel it better
1072 communicates which primitive refactorings it is composed of.
1074 For the metrics, I will at least measure the \emph{Coupling between object
1075 classes} (CBO) metric that is described by Chidamber and Kemerer in their
1076 article \emph{A Metrics Suite for Object Oriented
1077 Design}\citing{metricsSuite1994}.
1079 The project will then consist in implementing the \ExtractAndMoveMethod
1080 refactoring, as well as executing it over a larger code base. Then the effect of
1081 the change must be measured by calculating the chosen software metrics both
1082 before and after the execution. To be able to execute the refactoring
1083 automatically I have to make it analyze code to determine the best selections to
1084 extract into new methods.
1088 %\chapter{Planning the project}
1094 \chapter{The Project}
1096 \section{The problem statement}
1099 \section{Choosing the target language}
1100 Choosing which programming language the code that shall be manipulated shall be
1101 written in, is not a very difficult task. We choose to limit the possible
1102 languages to the object-oriented programming languages, since most of the
1103 terminology and literature regarding refactoring comes from the world of
1104 object-oriented programming. In addition, the language must have existing tool
1105 support for refactoring.
1107 The \emph{Java} programming language\footnote{\url{https://www.java.com/}} is
1108 the dominating language when it comes to example code in the literature of
1109 refactoring, and is thus a natural choice. Java is perhaps, currently the most
1110 influential programming language in the world, with its \emph{Java Virtual
1111 Machine} that runs on all of the most popular architectures and also supports
1112 dozens of other programming languages\footnote{They compile to java bytecode.},
1113 with \emph{Scala}, \emph{Clojure} and \emph{Groovy} as the most prominent ones.
1114 Java is currently the language that every other programming language is compared
1115 against. It is also the primary programming language for the author of this
1118 \section{Choosing the tools}
1119 When choosing a tool for manipulating Java, there are certain criterias that
1120 have to be met. First of all, the tool should have some existing refactoring
1121 support that this thesis can build upon. Secondly it should provide some kind of
1122 framework for parsing and analyzing Java source code. Third, it should itself be
1123 open source. This is both because of the need to be able to browse the code for
1124 the existing refactorings that is contained in the tool, and also because open
1125 source projects hold value in them selves. Another important aspect to consider
1126 is that open source projects of a certain size, usually has large communities of
1127 people connected to them, that are commited to answering questions regarding the
1128 use and misuse of the products, that to a large degree is made by the cummunity
1131 There is a certain class of tools that meet these criterias, namely the class of
1132 \emph{IDEs}\footnote{\emph{Integrated Development Environment}}. These are
1133 proagrams that is ment to support the whole production cycle of a cumputer
1134 program, and the most popular IDEs that support Java, generally have quite good
1135 refactoring support.
1137 The main contenders for this thesis is the \emph{Eclipse IDE}, with the
1138 \emph{Java development tools} (JDT), the \emph{IntelliJ IDEA Community Edition}
1139 and the \emph{NetBeans IDE} \see{toolSupport}. Eclipse and NetBeans are both
1140 free, open source and community driven, while the IntelliJ IDEA has an open
1141 sourced community edition that is free of charge, but also offer an
1142 \emph{Ultimate Edition} with an extended set of features, at additional cost.
1143 All three IDEs supports adding plugins to extend their functionality and tools
1144 that can be used to parse and analyze Java source code. But one of the IDEs
1145 stand out as a favorite, and that is the \emph{Eclipse IDE}. This is the most
1146 popular\citing{javaReport2011} among them and seems to be de facto standard IDE
1147 for Java development regardless of platform.
1149 \section{Organizing the project}
1150 All the parts of this master project is under version control with
1151 \emph{Git}\footnote{\url{http://git-scm.com/}}.
1153 The software written is organized as some Eclipse plugins. Writing a plugin is
1154 the natural way to utilize the API of Eclipse. This also makes it possible to
1155 provide a user interface to manually run operations on selections in program
1156 source code or whole projects/packages.
1158 When writing a plugin in Eclipse, one has access to resources such as the
1159 current workspace, the open editor and the current selection.
1161 \section{Continuous integration}
1162 The continuous integration server
1163 \emph{Jenkins}\footnote{\url{http://jenkins-ci.org/}} has been set up for the
1164 project\footnote{A work mostly done by the supervisor.}. It is used as a way to
1165 run tests and perform code coverage analysis.
1167 To be able to build the Eclipse plugins and run tests for them with Jenkins, the
1168 component assembly project
1169 \emph{Buckminster}\footnote{\url{http://www.eclipse.org/buckminster/}} is used,
1170 through its plugin for Jenkins. Buckminster provides for a way to specify the
1171 resources needed for building a project and where and how to find them.
1172 Buckminster also handles the setup of a target environment to run the tests in.
1173 All this is needed because the code to build depends on an Eclipse installation
1174 with various plugins.
1176 \subsection{Problems with AspectJ}
1177 The Buckminster build worked fine until introducing AspectJ into the project.
1178 When building projects using AspectJ, there are some additional steps that needs
1179 to be performed. First of all, the aspects themselves must be compiled. Then the
1180 aspects needs to be woven with the classes they affect. This demands a process
1181 that does multiple passes over the source code.
1183 When using AspectJ with Eclipse, the specialized compilation and the weaving can
1184 be handled by the \emph{AspectJ Development
1185 Tools}\footnote{\url{https://www.eclipse.org/ajdt/}}. This works all fine, but
1186 it complicates things when trying to build a project depending on Eclipse
1187 plugins outside of Eclipse. There is supposed to be a way to specify a compiler
1188 adapter for javac, together with the file extensions for the file types it shall
1189 operate. The AspectJ compiler adapter is called
1190 \typewithref{org.aspectj.tools.ant.taskdefs}{Ajc11CompilerAdapter}, and it works
1191 with files that has the extensions \code{*.java} and \code{*.aj}. I tried to
1192 setup this in the build properties file for the project containing the aspects,
1193 but to no avail. The project containing the aspects does not seem to be built at
1194 all, and the projects that depends on it complains that they cannot find certain
1197 I then managed to write an \emph{Ant}\footnote{\url{https://ant.apache.org/}}
1198 build file that utilizes the AspectJ compiler adapter, for the
1199 \code{no.uio.ifi.refaktor} plugin. The problem was then that it could no longer
1200 take advantage of the environment set up by Buckminster. The solution to this
1201 particular problem was of a ``hacky'' nature. It involves exporting the plugin
1202 dependencies for the project to an Ant build file, and copy the exported path
1203 into the existing build script. But then the Ant script needs to know where the
1204 local Eclipse installation is located. This is no problem when building on a
1205 local machine, but to utilize the setup done by Buckminster is a problem still
1206 unsolved. To get the classpath for the build setup correctly, and here comes the
1207 most ``hacky'' part of the solution, the Ant script has a target for copying the
1208 classpath elements into a directory relative to the project directory and
1209 checking it into Git. When no \code{ECLIPSE\_HOME} property is set while running
1210 Ant, the script uses the copied plugins instead of the ones provided by the
1211 Eclipse installation when building the project. This obviously creates some
1212 problems with maintaining the list of dependencies in the Ant file, as well as
1213 remembering to copy the plugins every time the list of dependencies change.
1215 The Ant script described above is run by Jenkins before the Buckminster setup
1216 and build. When setup like this, the Buckminster build succeeds for the projects
1217 not using AspectJ, and the tests are run as normal. This is all good, but it
1218 feels a little scary, since the reason for Buckminster not working with AspectJ
1221 The problems with building with AspectJ on the Jenkins server lasted for a
1222 while, before they were solved. This is reflected in the ``Test Result Trend''
1223 and ``Code Coverage Trend'' reported by Jenkins.
1226 \chapter{Refactorings in Eclipse JDT: Design, Shortcomings and Wishful
1227 Thinking}\label{ch:jdt_refactorings}
1229 This chapter will deal with some of the design behind refactoring support in
1230 Eclipse, and the JDT in specific. After which it will follow a section about
1231 shortcomings of the refactoring API in terms of composition of refactorings. The
1232 chapter will be concluded with a section telling some of the ways the
1233 implementation of refactorings in the JDT could have worked to facilitate
1234 composition of refactorings.
1237 The refactoring world of Eclipse can in general be separated into two parts: The
1238 language independent part and the part written for a specific programming
1239 language -- the language that is the target of the supported refactorings.
1240 \todo{What about the language specific part?}
1242 \subsection{The Language Toolkit}
1243 The Language Toolkit\footnote{The content of this section is a mixture of
1244 written material from
1245 \url{https://www.eclipse.org/articles/Article-LTK/ltk.html} and
1246 \url{http://www.eclipse.org/articles/article.php?file=Article-Unleashing-the-Power-of-Refactoring/index.html},
1247 the LTK source code and my own memory.}, or LTK for short, is the framework that
1248 is used to implement refactorings in Eclipse. It is language independent and
1249 provides the abstractions of a refactoring and the change it generates, in the
1250 form of the classes \typewithref{org.eclipse.ltk.core.refactoring}{Refactoring}
1251 and \typewithref{org.eclipse.ltk.core.refactoring}{Change}.
1253 There are also parts of the LTK that is concerned with user interaction, but
1254 they will not be discussed here, since they are of little value to us and our
1255 use of the framework. We are primarily interested in the parts that can be
1258 \subsubsection{The Refactoring Class}
1259 The abstract class \type{Refactoring} is the core of the LTK framework. Every
1260 refactoring that is going to be supported by the LTK have to end up creating an
1261 instance of one of its subclasses. The main responsibilities of subclasses of
1262 \type{Refactoring} is to implement template methods for condition checking
1263 (\methodwithref{org.eclipse.ltk.core.refactoring.Refactoring}{checkInitialConditions}
1265 \methodwithref{org.eclipse.ltk.core.refactoring.Refactoring}{checkFinalConditions}),
1267 \methodwithref{org.eclipse.ltk.core.refactoring.Refactoring}{createChange}
1268 method that creates and returns an instance of the \type{Change} class.
1270 If the refactoring shall support that others participate in it when it is
1271 executed, the refactoring has to be a processor-based
1272 refactoring\typeref{org.eclipse.ltk.core.refactoring.participants.ProcessorBasedRefactoring}.
1273 It then delegates to its given
1274 \typewithref{org.eclipse.ltk.core.refactoring.participants}{RefactoringProcessor}
1275 for condition checking and change creation. Participating in a refactoring can
1276 be useful in cases where the changes done to programming source code affects
1277 other related resources in the workspace. This can be names or paths in
1278 configuration files, or maybe one would like to perform additional logging of
1279 changes done in the workspace.
1281 \subsubsection{The Change Class}
1282 This class is the base class for objects that is responsible for performing the
1283 actual workspace transformations in a refactoring. The main responsibilities for
1284 its subclasses is to implement the
1285 \methodwithref{org.eclipse.ltk.core.refactoring.Change}{perform} and
1286 \methodwithref{org.eclipse.ltk.core.refactoring.Change}{isValid} methods. The
1287 \method{isValid} method verifies that the change object is valid and thus can be
1288 executed by calling its \method{perform} method. The \method{perform} method
1289 performs the desired change and returns an undo change that can be executed to
1290 reverse the effect of the transformation done by its originating change object.
1292 \subsubsection{Executing a Refactoring}\label{executing_refactoring}
1293 The life cycle of a refactoring generally follows two steps after creation:
1294 condition checking and change creation. By letting the refactoring object be
1296 \typewithref{org.eclipse.ltk.core.refactoring}{CheckConditionsOperation} that
1297 in turn is handled by a
1298 \typewithref{org.eclipse.ltk.core.refactoring}{CreateChangeOperation}, it is
1299 assured that the change creation process is managed in a proper manner.
1301 The actual execution of a change object has to follow a detailed life cycle.
1302 This life cycle is honored if the \type{CreateChangeOperation} is handled by a
1303 \typewithref{org.eclipse.ltk.core.refactoring}{PerformChangeOperation}. If also
1304 an undo manager\typeref{org.eclipse.ltk.core.refactoring.IUndoManager} is set
1305 for the \type{PerformChangeOperation}, the undo change is added into the undo
1308 \section{Shortcomings}
1309 This section is introduced naturally with a conclusion: The JDT refactoring
1310 implementation does not facilitate composition of refactorings.
1311 \todo{refine}This section will try to explain why, and also identify other
1312 shortcomings of both the usability and the readability of the JDT refactoring
1315 I will begin at the end and work my way toward the composition part of this
1318 \subsection{Absence of Generics in Eclipse Source Code}
1319 This section is not only concerning the JDT refactoring API, but also large
1320 quantities of the Eclipse source code. The code shows a striking absence of the
1321 Java language feature of generics. It is hard to read a class' interface when
1322 methods return objects or takes parameters of raw types such as \type{List} or
1323 \type{Map}. This sometimes results in having to read a lot of source code to
1324 understand what is going on, instead of relying on the available interfaces. In
1325 addition, it results in a lot of ugly code, making the use of typecasting more
1326 of a rule than an exception.
1328 \subsection{Composite Refactorings Will Not Appear as Atomic Actions}
1330 \subsubsection{Missing Flexibility from JDT Refactorings}
1331 The JDT refactorings are not made with composition of refactorings in mind. When
1332 a JDT refactoring is executed, it assumes that all conditions for it to be
1333 applied successfully can be found by reading source files that have been
1334 persisted to disk. They can only operate on the actual source material, and not
1335 (in-memory) copies thereof. This constitutes a major disadvantage when trying to
1336 compose refactorings, since if an exception occurs in the middle of a sequence
1337 of refactorings, it can leave the project in a state where the composite
1338 refactoring was only partially executed. It makes it hard to discard the changes
1339 done without monitoring and consulting the undo manager, an approach that is not
1342 \subsubsection{Broken Undo History}
1343 When designing a composed refactoring that is to be performed as a sequence of
1344 refactorings, you would like it to appear as a single change to the workspace.
1345 This implies that you would also like to be able to undo all the changes done by
1346 the refactoring in a single step. This is not the way it appears when a sequence
1347 of JDT refactorings is executed. It leaves the undo history filled up with
1348 individual undo actions corresponding to every single JDT refactoring in the
1349 sequence. This problem is not trivial to handle in Eclipse
1350 \see{hacking_undo_history}.
1352 \section{Wishful Thinking}
1355 \chapter{Composite Refactorings in Eclipse}
1357 \section{A Simple Ad Hoc Model}
1358 As pointed out in \myref{ch:jdt_refactorings}, the Eclipse JDT refactoring model
1359 is not very well suited for making composite refactorings. Therefore a simple
1360 model using changer objects (of type \type{RefaktorChanger}) is used as an
1361 abstraction layer on top of the existing Eclipse refactorings, instead of
1362 extending the \typewithref{org.eclipse.ltk.core.refactoring}{Refactoring} class.
1364 The use of an additional abstraction layer is a deliberate choice. It is due to
1365 the problem of creating a composite
1366 \typewithref{org.eclipse.ltk.core.refactoring}{Change} that can handle text
1367 changes that interfere with each other. Thus, a \type{RefaktorChanger} may, or
1368 may not, take advantage of one or more existing refactorings, but it is always
1369 intended to make a change to the workspace.
1371 \subsection{A typical \type{RefaktorChanger}}
1372 The typical refaktor changer class has two responsibilities, checking
1373 preconditions and executing the requested changes. This is not too different
1374 from the responsibilities of an LTK refactoring, with the distinction that a
1375 refaktor changer also executes the change, while an LTK refactoring is only
1376 responsible for creating the object that can later be used to do the job.
1378 Checking of preconditions is typically done by an
1379 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{Analyzer}. If the
1380 preconditions validate, the upcoming changes are executed by an
1381 \typewithref{no.uio.ifi.refaktor.change.executors}{Executor}.
1383 \section{The Extract and Move Method Refactoring}
1384 %The Extract and Move Method Refactoring is implemented mainly using these
1387 % \item \type{ExtractAndMoveMethodChanger}
1388 % \item \type{ExtractAndMoveMethodPrefixesExtractor}
1389 % \item \type{Prefix}
1390 % \item \type{PrefixSet}
1393 \subsection{The Building Blocks}
1394 This is a composite refactoring, and hence is built up using several primitive
1395 refactorings. These basic building blocks are, as its name implies, the
1396 \ExtractMethod refactoring\citing{refactoring} and the \MoveMethod
1397 refactoring\citing{refactoring}. In Eclipse, the implementations of these
1398 refactorings are found in the classes
1399 \typewithref{org.eclipse.jdt.internal.corext.refactoring.code}{ExtractMethodRefactoring}
1401 \typewithref{org.eclipse.jdt.internal.corext.refactoring.structure}{MoveInstanceMethodProcessor},
1402 where the last class is designed to be used together with the processor-based
1403 \typewithref{org.eclipse.ltk.core.refactoring.participants}{MoveRefactoring}.
1405 \subsubsection{The ExtractMethodRefactoring Class}
1406 This class is quite simple in its use. The only parameters it requires for
1407 construction is a compilation
1408 unit\typeref{org.eclipse.jdt.core.ICompilationUnit}, the offset into the source
1409 code where the extraction shall start, and the length of the source to be
1410 extracted. Then you have to set the method name for the new method together with
1411 its visibility and some not so interesting parameters.
1413 \subsubsection{The MoveInstanceMethodProcessor Class}
1414 For the Move Method, the processor requires a little more advanced input than
1415 the class for the Extract Method. For construction it requires a method
1416 handle\typeref{org.eclipse.jdt.core.IMethod} for the method that is to be moved.
1417 Then the target for the move have to be supplied as the variable binding from a
1418 chosen variable declaration. In addition to this, one have to set some
1419 parameters regarding setters/getters, as well as delegation.
1421 To make a working refactoring from the processor, one have to create a
1422 \type{MoveRefactoring} with it.
1424 \subsection{The ExtractAndMoveMethodChanger}
1426 The \typewithref{no.uio.ifi.refaktor.changers}{ExtractAndMoveMethodChanger}
1427 class is a subclass of the class
1428 \typewithref{no.uio.ifi.refaktor.changers}{RefaktorChanger}. It is responsible
1429 for analyzing and finding the best target for, and also executing, a composition
1430 of the Extract Method and Move Method refactorings. This particular changer is
1431 the one of my changers that is closest to being a true LTK refactoring. It can
1432 be reworked to be one if the problems with overlapping changes are resolved. The
1433 changer requires a text selection and the name of the new method, or else a
1434 method name will be generated. The selection has to be of the type
1435 \typewithref{no.uio.ifi.refaktor.utils}{CompilationUnitTextSelection}. This
1436 class is a custom extension to
1437 \typewithref{org.eclipse.jface.text}{TextSelection}, that in addition to the
1438 basic offset, length and similar methods, also carry an instance of the
1439 underlying compilation unit handle for the selection.
1442 \type{ExtractAndMoveMethodAnalyzer}}\label{extractAndMoveMethodAnalyzer}
1443 The analysis and precondition checking is done by the
1444 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{ExtractAnd\-MoveMethodAnalyzer}.
1445 First is check whether the selection is a valid selection or not, with respect
1446 to statement boundaries and that it actually contains any selections. Then it
1447 checks the legality of both extracting the selection and also moving it to
1448 another class. This checking of is performed by a range of checkers
1449 \see{checkers}. If the selection is approved as legal, it is analyzed to find
1450 the presumably best target to move the extracted method to.
1452 For finding the best suitable target the analyzer is using a
1453 \typewithref{no.uio.ifi.refaktor.analyze.collectors}{PrefixesCollector} that
1454 collects all the possible candidate targets for the refactoring. All the
1455 non-candidates is found by an
1456 \typewithref{no.uio.ifi.refaktor.analyze.collectors}{UnfixesCollector} that
1457 collects all the targets that will give some kind of error if used. (For
1458 details about the property collectors, se \myref{propertyCollectors}.) All
1459 prefixes (and unfixes) are represented by a
1460 \typewithref{no.uio.ifi.refaktor.extractors}{Prefix}, and they are collected
1461 into sets of prefixes. The safe prefixes is found by subtracting from the set of
1462 candidate prefixes the prefixes that is enclosing any of the unfixes. A prefix
1463 is enclosing an unfix if the unfix is in the set of its sub-prefixes. As an
1464 example, \texttt{``a.b''} is enclosing \texttt{``a''}, as is \texttt{``a''}. The
1465 safe prefixes is unified in a \type{PrefixSet}. If a prefix has only one
1466 occurrence, and is a simple expression, it is considered unsuitable as a move
1467 target. This occurs in statements such as \texttt{``a.foo()''}. For such
1468 statements it bares no meaning to extract and move them. It only generates an
1469 extra method and the calling of it.
1471 The most suitable target for the refactoring is found by finding the prefix with
1472 the most occurrences. If two prefixes have the same occurrence count, but they
1473 differ in length, the longest of them is chosen.
1475 \todoin{Clean up sections/subsections.}
1478 \type{ExtractAndMoveMethodExecutor}}\label{extractAndMoveMethodExecutor}
1479 If the analysis finds a possible target for the composite refactoring, it is
1481 \typewithref{no.uio.ifi.refaktor.change.executors}{ExtractAndMoveMethodExecutor}.
1482 It is composed of the two executors known as
1483 \typewithref{no.uio.ifi.refaktor.change.executors}{ExtractMethodRefactoringExecutor}
1485 \typewithref{no.uio.ifi.refaktor.change.executors}{MoveMethodRefactoringExecutor}.
1486 The \type{ExtractAndMoveMethodExecutor} is responsible for gluing the two
1487 together by feeding the \type{MoveMethod\-RefactoringExecutor} with the
1488 resources needed after executing the extract method refactoring
1489 \see{postExtractExecution}.
1491 \subsubsection{The \type{ExtractMethodRefactoringExecutor}}
1492 This executor is responsible for creating and executing an instance of the
1493 \type{ExtractMethodRefactoring} class. It is also responsible for collecting
1494 some post execution resources that can be used to find the method handle for the
1495 extracted method, as well as information about its parameters, including the
1496 variable they originated from.
1498 \subsubsection{The \type{MoveMethodRefactoringExecutor}}
1499 This executor is responsible for creating and executing an instance of the
1500 \type{MoveRefactoring}. The move refactoring is a processor-based refactoring,
1501 and for the Move Method refactoring it is the \type{MoveInstanceMethodProcessor}
1504 The handle for the method to be moved is found on the basis of the information
1505 gathered after the execution of the Extract Method refactoring. The only
1506 information the \type{ExtractMethodRefactoring} is sharing after its execution,
1507 regarding find the method handle, is the textual representation of the new
1508 method signature. Therefore it must be parsed, the strings for types of the
1509 parameters must be found and translated to a form that can be used to look up
1510 the method handle from its type handle. They have to be on the unresolved
1511 form.\todo{Elaborate?} The name for the type is found from the original
1512 selection, since an extracted method must end up in the same type as the
1515 When analyzing a selection prior to performing the Extract Method refactoring, a
1516 target is chosen. It has to be a variable binding, so it is either a field or a
1517 local variable/parameter. If the target is a field, it can be used with the
1518 \type{MoveInstanceMethodProcessor} as it is, since the extracted method still is
1519 in its scope. But if the target is local to the originating method, the target
1520 that is to be used for the processor must be among its parameters. Thus the
1521 target must be found among the extracted method's parameters. This is done by
1522 finding the parameter information object that corresponds to the parameter that
1523 was declared on basis of the original target's variable when the method was
1524 extracted. (The extracted method must take one such parameter for each local
1525 variable that is declared outside the selection that is extracted.) To match the
1526 original target with the correct parameter information object, the key for the
1527 information object is compared to the key from the original target's binding.
1528 The source code must then be parsed to find the method declaration for the
1529 extracted method. The new target must be found by searching through the
1530 parameters of the declaration and choose the one that has the same type as the
1531 old binding from the parameter information object, as well as the same name that
1532 is provided by the parameter information object.
1536 SearchBasedExtractAndMoveMethodChanger}\label{searchBasedExtractAndMoveMethodChanger}
1538 \typewithref{no.uio.ifi.refaktor.change.changers}{SearchBasedExtractAndMoveMethodChanger}
1539 is a changer whose purpose is to automatically analyze a method, and execute the
1540 \ExtractAndMoveMethod refactoring on it if it is a suitable candidate for the
1543 First, the \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{SearchBasedExtractAndMoveMethodAnalyzer} is used
1544 to analyze the method. If the method is found to be a candidate, the result from
1545 the analysis is fed to the \type{ExtractAndMoveMethodExecutor}, whose job is to
1546 execute the refactoring \see{extractAndMoveMethodExecutor}.
1548 \subsubsection{The SearchBasedExtractAndMoveMethodAnalyzer}
1549 This analyzer is responsible for analyzing all the possible text selections of a
1550 method and then choose the best result out of the analysis results that is, by
1551 the analyzer, considered to be the potential candidates for the Extract and Move
1554 Before the analyzer is able to work with the text selections of a method, it
1555 needs to generate them. To do this, it parses the method to obtain a
1556 \type{MethodDeclaration} for it \see{astEclipse}. Then there is a statement
1557 lists creator that creates statements lists of the different groups of
1558 statements in the body of the method declaration. A text selections generator
1559 generates text selections of all the statement lists for the analyzer to work
1562 \paragraph{The statement lists creator}
1563 is responsible for generating lists of statements for all the possible levels of
1564 statements in the method. The statement lists creator is implemented as an AST
1565 visitor \see{astVisitor}. It generates lists of statements by visiting all the
1566 blocks in the method declaration and stores their statements in a collection of
1567 statement lists. In addition, it visits all of the other statements that can
1568 have a statement as a child, such as the different control structures and the
1571 The switch statement is the only kind of statement that is not straight forward
1572 to obtain the child statements from. It stores all of its children in a flat
1573 list. Its switch case statements are included in this list. This means that
1574 there are potential statement lists between all of these case statements. The
1575 list of statements from a switch statement is therefore traversed, and the
1576 statements between the case statements are grouped as separate lists.
1578 There is an example of how the statement lists creator would generate lists for
1579 a simple method in \myref{lst:statementListsExample}.
1582 \def\charwidth{5.7pt}
1583 \def\indent{4*\charwidth}
1584 \def\lineheight{\baselineskip}
1585 \def\mintedtop{\lineheight}
1587 \begin{tikzpicture}[overlay, yscale=-1]
1588 \tikzstyle{overlaybox}=[fill=lightgray,opacity=0.2]
1589 \draw[overlaybox] (0,\mintedtop+\lineheight) rectangle
1590 +(22*\charwidth,10*\lineheight);
1591 \draw[overlaybox] (\indent,\mintedtop+2*\lineheight) rectangle
1592 +(13*\charwidth,\lineheight);
1593 \draw[overlaybox] (2*\indent,\mintedtop+6*\lineheight) rectangle
1594 +(13*\charwidth,2*\lineheight);
1595 \draw[overlaybox] (2*\indent,\mintedtop+9*\lineheight) rectangle
1596 +(13*\charwidth,\lineheight);
1598 \begin{minted}{java}
1612 \caption{Example of how the statement lists creator would group a simple method
1613 into lists of statements. Each highlighted rectangle represents a list.}
1614 \label{lst:statementListsExample}
1617 \paragraph{The text selections generator} generates text selections for each
1618 list of statements from the statement lists creator. Conceptually, the generator
1619 generates a text selection for every possible ordered \todo{make clearer}
1620 combination of statements in a list. For a list of statements, the boundary
1621 statements span out a text selection. This means that there are many different
1622 lists that could span out the same selection.
1624 In practice, the text selections are calculated by only one traversal of the
1625 statement list. There is a set of generated text selections. For each statement,
1626 there is created a temporary set of selections, in addition to a text selection
1627 based on the offset and length of the statement. This text selection is added to
1628 the temporary set. Then the new selection is added with every selection from the
1629 set of generated text selections. These new selections are added to the
1630 temporary set. Then the temporary set of selections is added to the set of
1631 generated text selections. The result of adding two text selections is a new
1632 text selection spanned out by the two addends.
1634 \paragraph{Finding the candidate} for the refactoring is done by analyzing all
1635 the generated text selection with the \type{ExtractAndMoveMethodAnalyzer}
1636 \see{extractAndMoveMethodAnalyzer}. If the analyzer generates a useful result,
1637 an \type{ExtractAndMoveMethodCandidate} is created from it, that is kept in a
1638 list of potential candidates. If no candidates are found, the
1639 \type{NoTargetFoundException} is thrown.
1641 Since only one of the candidates can be chosen, the analyzer must sort out which
1642 candidate to choose. The sorting is done by the static \method{sort} method of
1643 \type{Collections}. The comparison in this sorting is done by an
1644 \type{ExtractAndMoveMethodCandidateComparator}.
1645 \todoin{Write about the
1646 ExtractAndMoveMethodCandidateComparator/FavorNoUnfixesCandidateComparator}
1648 \paragraph{The complexity} of how many text selections that needs to be analyzed
1649 for a total of $n$ statements is bounded by $O(n^2)$.
1652 The number of text selections that need to be analyzed for each list of
1653 statements of length $n$, is exactly
1656 \sum_{i=1}^{n} i = \frac{n(n+1)}{2}
1657 \label{eq:complexityStatementList}
1659 \label{thm:numberOfTextSelection}
1663 For $n=1$ this is trivial: $\frac{1(1+1)}{2} = \frac{2}{2} = 1$. One statement
1664 equals one selection.
1666 For $n=2$, you get one text selection for the first statement. For the second,
1667 you get one selection for the statement itself, and one selection for the two
1668 of them combined. This equals three selections. $\frac{2(2+1)}{2} =
1671 For $n=3$, you get 3 selections for the two first statements, as in the case
1672 where $n=2$. In addition you get one selection for the third statement itself,
1673 and two more statements for the combinations of it with the two previous
1674 statements. This equals six selections. $\frac{3(3+1)}{2} = \frac{12}{2} = 6$.
1676 Assume that for $n=k$ there exists $\frac{k(k+1)}{2}$ text selections. Then we
1677 want to add selections for another statement, following the previous $k$
1678 statements. So, for $n=k+1$, we get one additional selection for the statement
1679 itself. Then we get one selection for each pair of the new selection and the
1680 previous $k$ statements. So the total number of selections will be the number
1681 of already generated selections, plus $k$ for every pair, plus one for the
1682 statement itself: $\frac{k(k+1)}{2} + k +
1683 1 = \frac{k(k+1)+2k+2}{2} = \frac{k(k+1)+2(k+1)}{2} = \frac{(k+1)(k+2)}{2} =
1684 \frac{(k+1)((k+1)+1)}{2} = \sum_{i=1}^{k+1} i$
1688 The number of text selections for a body of statements is maximized if all the
1689 statements are at the same level.
1690 \label{thm:textSelectionsMaximized}
1694 Assume we have a body of, in total, $k$ stamements. Let
1695 $l,\cdots,m,(k-l-\cdots-m)$ be the lengths of the lists of statements in the
1696 body, with $l+\cdots+m<k \Rightarrow l,\cdots,m<k$.
1698 Then, the number of text selections that are generated for the $k$ statements
1704 \frac{(k-l-\cdots-m)((k-l-\cdots-m)+ 1)}{2} + \frac{l(l+1)}{2} + \cdots +
1705 \frac{m(m+1)}{2} = \\
1706 \frac{k^2 - 2kl - \cdots - 2km + l^2 + \cdots + m^2 + k - l - \cdots - m}{2}
1707 + \frac{l^2+l}{2} + \cdots + \frac{m^2+m}{2} = \\
1708 \frac{k^2 + k + 2l^2 - 2kl + \cdots + 2m^2 - 2km}{2}
1712 It then remains to show that this inequality holds:
1715 \frac{k^2 + k + 2l^2 - 2kl + \cdots + 2m^2 - 2km}{2} < \frac{k(k+1)}{2} =
1719 By multiplication by $2$ on both sides, and by removing the equal parts, we get
1722 2l^2 - 2kl + \cdots + 2m^2 - 2km < 0
1725 Since $l,\cdots,m<k$, we have that $\forall i \in \{l,\cdots,m\} : 2ki > 2i^2$,
1726 so all the pairs of parts on the form $2i^2-2ki$ are negative. In sum, the
1731 Therefore, the complexity for the number of selections that needs to be analyzed
1732 for a body of $n$ statements is $O\bigl(\frac{n(n+1)}{2}\bigr) = O(n^2)$.
1735 \subsection{Finding the IMethod}\label{postExtractExecution}
1736 \todoin{Rename section. Write??}
1739 \subsection{The Prefix Class}
1740 This class exists mainly for holding data about a prefix, such as the expression
1741 that the prefix represents and the occurrence count of the prefix within a
1742 selection. In addition to this, it has some functionality such as calculating
1743 its sub-prefixes and intersecting it with another prefix. The definition of the
1744 intersection between two prefixes is a prefix representing the longest common
1745 expression between the two.
1747 \subsection{The PrefixSet Class}
1748 A prefix set holds elements of type \type{Prefix}. It is implemented with the
1749 help of a \typewithref{java.util}{HashMap} and contains some typical set
1750 operations, but it does not implement the \typewithref{java.util}{Set}
1751 interface, since the prefix set does not need all of the functionality a
1752 \type{Set} requires to be implemented. In addition It needs some other
1753 functionality not found in the \type{Set} interface. So due to the relatively
1754 limited use of prefix sets, and that it almost always needs to be referenced as
1755 such, and not a \type{Set<Prefix>}, it remains as an ad hoc solution to a
1758 There are two ways adding prefixes to a \type{PrefixSet}. The first is through
1759 its \method{add} method. This works like one would expect from a set. It adds
1760 the prefix to the set if it does not already contain the prefix. The other way
1761 is to \emph{register} the prefix with the set. When registering a prefix, if the
1762 set does not contain the prefix, it is just added. If the set contains the
1763 prefix, its count gets incremented. This is how the occurrence count is handled.
1765 The prefix set also computes the set of prefixes that is not enclosing any
1766 prefixes of another set. This is kind of a set difference operation only for
1769 \subsection{Hacking the Refactoring Undo
1770 History}\label{hacking_undo_history}
1771 \todoin{Where to put this section?}
1773 As an attempt to make multiple subsequent changes to the workspace appear as a
1774 single action (i.e. make the undo changes appear as such), I tried to alter
1775 the undo changes\typeref{org.eclipse.ltk.core.refactoring.Change} in the history
1776 of the refactorings.
1778 My first impulse was to remove the, in this case, last two undo changes from the
1779 undo manager\typeref{org.eclipse.ltk.core.refactoring.IUndoManager} for the
1780 Eclipse refactorings, and then add them to a composite
1781 change\typeref{org.eclipse.ltk.core.refactoring.CompositeChange} that could be
1782 added back to the manager. The interface of the undo manager does not offer a
1783 way to remove/pop the last added undo change, so a possible solution could be to
1784 decorate\citing{designPatterns} the undo manager, to intercept and collect the
1785 undo changes before delegating to the \method{addUndo}
1786 method\methodref{org.eclipse.ltk.core.refactoring.IUndoManager}{addUndo} of the
1787 manager. Instead of giving it the intended undo change, a null change could be
1788 given to prevent it from making any changes if run. Then one could let the
1789 collected undo changes form a composite change to be added to the manager.
1791 There is a technical challenge with this approach, and it relates to the undo
1792 manager, and the concrete implementation
1793 UndoManager2\typeref{org.eclipse.ltk.internal.core.refactoring.UndoManager2}.
1794 This implementation is designed in a way that it is not possible to just add an
1795 undo change, you have to do it in the context of an active
1796 operation\typeref{org.eclipse.core.commands.operations.TriggeredOperations}.
1797 One could imagine that it might be possible to trick the undo manager into
1798 believing that you are doing a real change, by executing a refactoring that is
1799 returning a kind of null change that is returning our composite change of undo
1800 refactorings when it is performed.
1802 Apart from the technical problems with this solution, there is a functional
1803 problem: If it all had worked out as planned, this would leave the undo history
1804 in a dirty state, with multiple empty undo operations corresponding to each of
1805 the sequentially executed refactoring operations, followed by a composite undo
1806 change corresponding to an empty change of the workspace for rounding of our
1807 composite refactoring. The solution to this particular problem could be to
1808 intercept the registration of the intermediate changes in the undo manager, and
1809 only register the last empty change.
1811 Unfortunately, not everything works as desired with this solution. The grouping
1812 of the undo changes into the composite change does not make the undo operation
1813 appear as an atomic operation. The undo operation is still split up into
1814 separate undo actions, corresponding to the change done by its originating
1815 refactoring. And in addition, the undo actions has to be performed separate in
1816 all the editors involved. This makes it no solution at all, but a step toward
1819 There might be a solution to this problem, but it remains to be found. The
1820 design of the refactoring undo management is partly to be blamed for this, as it
1821 it is to complex to be easily manipulated.
1826 \chapter{Analyzing Source Code in Eclipse}
1828 \section{The Java model}\label{javaModel}
1829 The Java model of Eclipse is its internal representation of a Java project. It
1830 is light-weight, and has only limited possibilities for manipulating source
1831 code. It is typically used as a basis for the Package Explorer in Eclipse.
1833 The elements of the Java model is only handles to the underlying elements. This
1834 means that the underlying element of a handle does not need to actually exist.
1835 Hence the user of a handle must always check that it exist by calling the
1836 \method{exists} method of the handle.
1838 The handles with descriptions is listed in \myref{tab:javaModel}.
1843 \newcolumntype{L}[1]{>{\hsize=#1\hsize\raggedright\arraybackslash}X}%
1844 % sum must equal number of columns (3)
1845 \begin{tabularx}{\textwidth}{| L{0.7} | L{1.1} | L{1.2} |}
1847 \textbf{Project Element} & \textbf{Java Model element} &
1848 \textbf{Description} \\
1850 Java project & \type{IJavaProject} & The Java project which contains all other objects. \\
1852 Source folder /\linebreak[2] binary folder /\linebreak[3] external library &
1853 \type{IPackageFragmentRoot} & Hold source or binary files, can be a folder
1854 or a library (zip / jar file). \\
1856 Each package & \type{IPackageFragment} & Each package is below the
1857 \type{IPackageFragmentRoot}, sub-packages are not leaves of the package,
1858 they are listed directed under \type{IPackageFragmentRoot}. \\
1860 Java Source file & \type{ICompilationUnit} & The Source file is always below
1861 the package node. \\
1863 Types /\linebreak[2] Fields /\linebreak[3] Methods & \type{IType} /
1865 \type{IField} /\linebreak[3] \type{IMethod} & Types, fields and methods. \\
1868 \caption{The elements of the Java Model. {\footnotesize Taken from
1869 \url{http://www.vogella.com/tutorials/EclipseJDT/article.html}}}
1870 \label{tab:javaModel}
1873 The hierarchy of the Java Model is shown in \myref{fig:javaModel}.
1877 \begin{tikzpicture}[%
1878 grow via three points={one child at (0,-0.7) and
1879 two children at (0,-0.7) and (0,-1.4)},
1880 edge from parent path={(\tikzparentnode.south west)+(0.5,0) |-
1881 (\tikzchildnode.west)}]
1882 \tikzstyle{every node}=[draw=black,thick,anchor=west]
1883 \tikzstyle{selected}=[draw=red,fill=red!30]
1884 \tikzstyle{optional}=[dashed,fill=gray!50]
1885 \node {\type{IJavaProject}}
1886 child { node {\type{IPackageFragmentRoot}}
1887 child { node {\type{IPackageFragment}}
1888 child { node {\type{ICompilationUnit}}
1889 child { node {\type{IType}}
1890 child { node {\type{\{ IType \}*}}
1891 child { node {\type{\ldots}}}
1894 child { node {\type{\{ IField \}*}}}
1895 child { node {\type{IMethod}}
1896 child { node {\type{\{ IType \}*}}
1897 child { node {\type{\ldots}}}
1902 child { node {\type{\{ IMethod \}*}}}
1911 child { node {\type{\{ IType \}*}}}
1922 child { node {\type{\{ ICompilationUnit \}*}}}
1935 child { node {\type{\{ IPackageFragment \}*}}}
1950 child { node {\type{\{ IPackageFragmentRoot \}*}}}
1953 \caption{The Java model of Eclipse. ``\type{\{ SomeElement \}*}'' means
1954 \type{SomeElement} zero or more times. For recursive structures,
1955 ``\type{\ldots}'' is used.}
1956 \label{fig:javaModel}
1959 \section{The Abstract Synax Tree}
1960 Eclipse is following the common paradigm of using an abstract syntaxt tree for
1961 source code analysis and manipulation.
1963 When parsing program source code into something that can be used as a foundation
1964 for analysis, the start of the process follows the same steps as in a compiler.
1965 This is all natural, because the way a compiler anayzes code is no different
1966 from how source manipulation programs would do it, except for some properties of
1967 code that is analyzed in the parser, and that they may be differing in what
1968 kinds of properties they analyze. Thus the process of translation source code
1969 into a structure that is suitable for analyzing, can be seen as a kind of
1970 interrupted compilation process \see{fig:interruptedCompilationProcess}.
1975 base/.style={anchor=north, align=center, rectangle, minimum height=1.4cm},
1976 basewithshadow/.style={base, drop shadow, fill=white},
1977 outlined/.style={basewithshadow, draw, rounded corners, minimum
1979 primary/.style={outlined, font=\bfseries},
1980 dashedbox/.style={outlined, dashed},
1981 arrowpath/.style={black, align=center, font=\small},
1982 processarrow/.style={arrowpath, ->, >=angle 90, shorten >=1pt},
1984 \begin{tikzpicture}[node distance=1.3cm and 3cm, scale=1, every
1985 node/.style={transform shape}]
1986 \node[base](AuxNode1){\small source code};
1987 \node[primary, right=of AuxNode1, xshift=-2.5cm](Scanner){Scanner};
1988 \node[primary, right=of Scanner, xshift=0.5cm](Parser){Parser};
1989 \node[dashedbox, below=of Parser](SemanticAnalyzer){Semantic\\Analyzer};
1990 \node[dashedbox, left=of SemanticAnalyzer](SourceCodeOptimizer){Source
1992 \node[dashedbox, below=of SourceCodeOptimizer
1993 ](CodeGenerator){Code\\Generator};
1994 \node[dashedbox, right=of CodeGenerator](TargetCodeOptimizer){Target
1996 \node[base, right=of TargetCodeOptimizer](AuxNode2){};
1998 \draw[processarrow](AuxNode1) -- (Scanner);
2000 \path[arrowpath] (Scanner) -- node [sloped](tokens){tokens}(Parser);
2001 \draw[processarrow](Scanner) -- (tokens) -- (Parser);
2003 \path[arrowpath] (Parser) -- node (syntax){syntax
2004 tree}(SemanticAnalyzer);
2005 \draw[processarrow](Parser) -- (syntax) -- (SemanticAnalyzer);
2007 \path[arrowpath] (SemanticAnalyzer) -- node
2008 [sloped](annotated){annotated\\tree}(SourceCodeOptimizer);
2009 \draw[processarrow, dashed](SemanticAnalyzer) -- (annotated) --
2010 (SourceCodeOptimizer);
2012 \path[arrowpath] (SourceCodeOptimizer) -- node
2013 (intermediate){intermediate code}(CodeGenerator);
2014 \draw[processarrow, dashed](SourceCodeOptimizer) -- (intermediate) --
2017 \path[arrowpath] (CodeGenerator) -- node [sloped](target1){target
2018 code}(TargetCodeOptimizer);
2019 \draw[processarrow, dashed](CodeGenerator) -- (target1) --
2020 (TargetCodeOptimizer);
2022 \path[arrowpath](TargetCodeOptimizer) -- node [sloped](target2){target
2024 \draw[processarrow, dashed](TargetCodeOptimizer) -- (target2) (AuxNode2);
2026 \caption{Interrupted compilation process. {\footnotesize (Full compilation
2027 process borrowed from \emph{Compiler construction: principles and practice}
2028 by Kenneth C. Louden\citing{louden1997}.)}}
2029 \label{fig:interruptedCompilationProcess}
2032 The process starts with a \emph{scanner}, or lexer. The job of the scanner is to
2033 read the source code and divide it into tokens for the parser. Therefore, it is
2034 also sometimes called a tokenizer. A token is a logical unit, defined in the
2035 language specification, consisting of one or more consecutive characters. In
2036 the java language the tokens can for instance be the \var{this} keyword, a curly
2037 bracket \var{\{} or a \var{nameToken}. It is recognized by the scanner on the
2038 basis of something eqivalent of a regular expression. This part of the process
2039 is often implemented with the use of a finite automata. In fact, it is common to
2040 specify the tokens in regular expressions, that in turn is translated into a
2041 finite automata lexer. This process can be automated.
2043 The program component used to translate a a stream of tokens into something
2044 meaningful, is called a parser. A parser is fed tokens from the scanner and
2045 performs an analysis of the structure of a program. It verifies that the syntax
2046 is correct according to the grammar rules of a language, that is usually
2047 specified in a context-free grammar, and often in a variant of the
2049 Form}\footnote{\url{https://en.wikipedia.org/wiki/Backus-Naur\_Form}}. The
2050 result coming from the parser is in the form of an \emph{Abstract Syntax Tree},
2051 AST for short. It is called \emph{abstract}, because the structure does not
2052 contain all of the tokens produced by the scanner. It only contain logical
2053 constructs, and because it forms a tree, all kinds of parentheses and brackets
2054 are implicit in the structure. It is this AST that is used when performing the
2055 semantic analysis of the code.
2057 As an example we can think of the expression \code{(5 + 7) * 2}. The root of
2058 this tree would in Eclipse be an \type{InfixExpression} with the operator
2059 \var{TIMES}, and a left operand that is also an \type{InfixExpression} with the
2060 operator \var{PLUS}. The left operand \type{InfixExpression}, has in turn a left
2061 operand of type \type{NumberLiteral} with the value \var{``5''} and a right
2062 operand \type{NumberLiteral} with the value \var{``7''}. The root will have a
2063 right operand of type \type{NumberLiteral} and value \var{``2''}. The AST for
2064 this expression is illustrated in \myref{fig:astInfixExpression}.
2066 Contrary to the Java Model, an abstract syntaxt tree is a heavy-weight
2067 representation of source code. It contains information about propertes like type
2068 bindings for variables and variable bindings for names.
2073 \begin{tikzpicture}[scale=0.8]
2074 \tikzset{level distance=40pt}
2075 \tikzset{sibling distance=5pt}
2076 \tikzstyle{thescale}=[scale=0.8]
2077 \tikzset{every tree node/.style={align=center}}
2078 \tikzset{edge from parent/.append style={thick}}
2079 \tikzstyle{inode}=[rectangle,rounded corners,draw,fill=lightgray,drop
2080 shadow,align=center]
2081 \tikzset{every internal node/.style={inode}}
2082 \tikzset{every leaf node/.style={draw=none,fill=none}}
2084 \Tree [.\type{InfixExpression} [.\type{InfixExpression}
2085 [.\type{NumberLiteral} \var{``5''} ] [.\type{Operator} \var{PLUS} ]
2086 [.\type{NumberLiteral} \var{``7''} ] ]
2087 [.\type{Operator} \var{TIMES} ]
2088 [.\type{NumberLiteral} \var{``2''} ]
2091 \caption{The abstract syntax tree for the expression \code{(5 + 7) * 2}.}
2092 \label{fig:astInfixExpression}
2095 \subsection{The AST in Eclipse}\label{astEclipse}
2096 In Eclipse, every node in the AST is a child of the abstract superclass
2097 \typewithref{org.eclipse.jdt.core.dom}{ASTNode}. Every \type{ASTNode}, among a
2098 lot of other things, provides information about its position and length in the
2099 source code, as well as a reference to its parent and to the root of the tree.
2101 The root of the AST is always of type \type{CompilationUnit}. It is not the same
2102 as an instance of an \type{ICompilationUnit}, which is the compilation unit
2103 handle of the Java model. The children of a \type{CompilationUnit} is an
2104 optional \type{PackageDeclaration}, zero or more nodes of type
2105 \type{ImportDecaration} and all its top-level type declarations that has node
2106 types \type{AbstractTypeDeclaration}.
2108 An \type{AbstractType\-Declaration} can be one of the types
2109 \type{AnnotationType\-Declaration}, \type{Enum\-Declaration} or
2110 \type{Type\-Declaration}. The children of an \type{AbstractType\-Declaration}
2111 must be a subtype of a \type{BodyDeclaration}. These subtypes are:
2112 \type{AnnotationTypeMember\-Declaration}, \type{EnumConstant\-Declaration},
2113 \type{Field\-Declaration}, \type{Initializer} and \type{Method\-Declaration}.
2115 Of the body declarations, the \type{Method\-Declaration} is the most interesting
2116 one. Its children include lists of modifiers, type parameters, parameters and
2117 exceptions. It has a return type node and a body node. The body, if present, is
2118 of type \type{Block}. A \type{Block} is itself a \type{Statement}, and its
2119 children is a list of \type{Statement} nodes.
2121 There are too many types of the abstract type \type{Statement} to list up, but
2122 there exists a subtype of \type{Statement} for every statement type of Java, as
2123 one would expect. This also applies to the abstract type \type{Expression}.
2124 However, the expression \type{Name} is a little special, since it is both used
2125 as an operand in compound expressions, as well as for names in type declarations
2128 There is an overview of some of the structure of an Eclipse AST in
2129 \myref{fig:astEclipse}.
2133 \begin{tikzpicture}[scale=0.8]
2134 \tikzset{level distance=50pt}
2135 \tikzset{sibling distance=5pt}
2136 \tikzstyle{thescale}=[scale=0.8]
2137 \tikzset{every tree node/.style={align=center}}
2138 \tikzset{edge from parent/.append style={thick}}
2139 \tikzstyle{inode}=[rectangle,rounded corners,draw,fill=lightgray,drop
2140 shadow,align=center]
2141 \tikzset{every internal node/.style={inode}}
2142 \tikzset{every leaf node/.style={draw=none,fill=none}}
2144 \Tree [.\type{CompilationUnit} [.\type{[ PackageDeclaration ]} [.\type{Name} ]
2145 [.\type{\{ Annotation \}*} ] ]
2146 [.\type{\{ ImportDeclaration \}*} [.\type{Name} ] ]
2147 [.\type{\{ AbstractTypeDeclaration \}+} [.\node(site){\type{\{
2148 BodyDeclaration \}*}}; ] [.\type{SimpleName} ] ]
2150 \begin{scope}[shift={(0.5,-6)}]
2151 \node[inode,thescale](root){\type{MethodDeclaration}};
2152 \node[inode,thescale](modifiers) at (4.5,-5){\type{\{ IExtendedModifier \}*}
2153 \\ {\footnotesize (Of type \type{Modifier} or \type{Annotation})}};
2154 \node[inode,thescale](typeParameters) at (-6,-3.5){\type{\{ TypeParameter
2156 \node[inode,thescale](parameters) at (-5,-5){\type{\{
2157 SingleVariableDeclaration \}*} \\ {\footnotesize (Parameters)}};
2158 \node[inode,thescale](exceptions) at (5,-3){\type{\{ Name \}*} \\
2159 {\footnotesize (Exceptions)}};
2160 \node[inode,thescale](return) at (-6.5,-2){\type{Type} \\ {\footnotesize
2162 \begin{scope}[shift={(0,-5)}]
2163 \Tree [.\node(body){\type{[ Block ]} \\ {\footnotesize (Body)}};
2164 [.\type{\{ Statement \}*} [.\type{\{ Expression \}*} ]
2165 [.\type{\{ Statement \}*} [.\type{\ldots} ]]
2170 \draw[->,>=triangle 90,shorten >=1pt](root.east)..controls +(east:2) and
2171 +(south:1)..(site.south);
2173 \draw (root.south) -- (modifiers);
2174 \draw (root.south) -- (typeParameters);
2175 \draw (root.south) -- ($ (parameters.north) + (2,0) $);
2176 \draw (root.south) -- (exceptions);
2177 \draw (root.south) -- (return);
2178 \draw (root.south) -- (body);
2181 \caption{The format of the abstract syntax tree in Eclipse.}
2182 \label{fig:astEclipse}
2184 \todoin{Add more to the AST format tree? \myref{fig:astEclipse}}
2186 \section{The ASTVisitor}\label{astVisitor}
2187 So far, the only thing that has been adressed is how the the data that is going
2188 to be the basis for our analysis is structured. Another aspect of it is how we
2189 are going to traverse the AST to gather the information we need, so we can
2190 conclude about the properties we are analysing. It is of course possible to
2191 start at the top of the tree, and manually search through its nodes for the ones
2192 we are looking for, but that is a bit inconvenient. To be able to efficiently
2193 utilize such an approach, we would need to make our own framework for traversing
2194 the tree and visiting only the types of nodes we are after. Luckily, this
2195 functionality is already provided in Eclipse, by its
2196 \typewithref{org.eclipse.jdt.core.dom}{ASTVisitor}.
2198 The Eclipse AST, together with its \type{ASTVisitor}, follows the \emph{Visitor}
2199 pattern\citing{designPatterns}. The intent of this design pattern is to
2200 facilitate extending the functionality of classes without touching the classes
2203 Let us say that there is a class hierarchy of \emph{Elements}. These elements
2204 all have a method \method{accept(Visitor visitor)}. In its simplest form, the
2205 \method{accept} method just calls the \method{visit} method of the visitor with
2206 itself as an argument, like this: \code{visitor.visit(this)}. For the visitors
2207 to be able to extend the functionality of all the classes in the elements
2208 hierarchy, each \type{Visitor} must have one visit method for each concrete
2209 class in the hierarchy. Say the hierarchy consists of the concrete classes
2210 \type{ConcreteElementA} and \type{ConcreteElementB}. Then each visitor must have
2211 the (possibly empty) methods \method{visit(ConcreteElementA element)} and
2212 \method{visit(ConcreteElementB element)}. This scenario is depicted in
2213 \myref{fig:visitorPattern}.
2217 \tikzstyle{abstract}=[rectangle, draw=black, fill=white, drop shadow, text
2218 centered, anchor=north, text=black, text width=6cm, every one node
2219 part/.style={align=center, font=\bfseries\itshape}]
2220 \tikzstyle{concrete}=[rectangle, draw=black, fill=white, drop shadow, text
2221 centered, anchor=north, text=black, text width=6cm]
2222 \tikzstyle{inheritarrow}=[->, >=open triangle 90, thick]
2223 \tikzstyle{commentarrow}=[->, >=angle 90, dashed]
2224 \tikzstyle{line}=[-, thick]
2225 \tikzset{every one node part/.style={align=center, font=\bfseries}}
2226 \tikzset{every second node part/.style={align=center, font=\ttfamily}}
2228 \begin{tikzpicture}[node distance=1cm, scale=0.8, every node/.style={transform
2230 \node (Element) [abstract, rectangle split, rectangle split parts=2]
2232 \nodepart{one}{Element}
2233 \nodepart{second}{+accept(visitor: Visitor)}
2235 \node (AuxNode01) [text width=0, minimum height=2cm, below=of Element] {};
2236 \node (ConcreteElementA) [concrete, rectangle split, rectangle split
2237 parts=2, left=of AuxNode01]
2239 \nodepart{one}{ConcreteElementA}
2240 \nodepart{second}{+accept(visitor: Visitor)}
2242 \node (ConcreteElementB) [concrete, rectangle split, rectangle split
2243 parts=2, right=of AuxNode01]
2245 \nodepart{one}{ConcreteElementB}
2246 \nodepart{second}{+accept(visitor: Visitor)}
2249 \node[comment, below=of ConcreteElementA] (CommentA) {visitor.visit(this)};
2251 \node[comment, below=of ConcreteElementB] (CommentB) {visitor.visit(this)};
2253 \node (AuxNodeX) [text width=0, minimum height=1cm, below=of AuxNode01] {};
2255 \node (Visitor) [abstract, rectangle split, rectangle split parts=2,
2258 \nodepart{one}{Visitor}
2259 \nodepart{second}{+visit(ConcreteElementA)\\+visit(ConcreteElementB)}
2261 \node (AuxNode02) [text width=0, minimum height=2cm, below=of Visitor] {};
2262 \node (ConcreteVisitor1) [concrete, rectangle split, rectangle split
2263 parts=2, left=of AuxNode02]
2265 \nodepart{one}{ConcreteVisitor1}
2266 \nodepart{second}{+visit(ConcreteElementA)\\+visit(ConcreteElementB)}
2268 \node (ConcreteVisitor2) [concrete, rectangle split, rectangle split
2269 parts=2, right=of AuxNode02]
2271 \nodepart{one}{ConcreteVisitor2}
2272 \nodepart{second}{+visit(ConcreteElementA)\\+visit(ConcreteElementB)}
2276 \draw[inheritarrow] (ConcreteElementA.north) -- ++(0,0.7) -|
2278 \draw[line] (ConcreteElementA.north) -- ++(0,0.7) -|
2279 (ConcreteElementB.north);
2281 \draw[inheritarrow] (ConcreteVisitor1.north) -- ++(0,0.7) -|
2283 \draw[line] (ConcreteVisitor1.north) -- ++(0,0.7) -|
2284 (ConcreteVisitor2.north);
2286 \draw[commentarrow] (CommentA.north) -- (ConcreteElementA.south);
2287 \draw[commentarrow] (CommentB.north) -- (ConcreteElementB.south);
2291 \caption{The Visitor Pattern.}
2292 \label{fig:visitorPattern}
2295 The use of the visitor pattern can be appropriate when the hierarchy of elements
2296 is mostly stable, but the family of operations over its elements is constantly
2297 growing. This is clearly the cas for the Eclipse AST, since the hierarchy of
2298 type \type{ASTNode} is very stable, but the functionality of its elements is
2299 extended every time someone needs to operate on the AST. Another aspect of the
2300 Eclipse implementation is that it is a public API, and the visitor pattern is an
2301 easy way to provide access to the nodes in the tree.
2303 The version of the visitor pattern implemented for the AST nodes in Eclipse also
2304 provides an elegant way to traverse the tree. It does so by following the
2305 convention that every node in the tree first let the visitor visit itself,
2306 before it also makes all its children accept the visitor. The children are only
2307 visited if the visit method of their parent returns \var{true}. This pattern
2308 then makes for a prefix traversal of the AST. If postfix traversal is desired,
2309 the visitors also has \method{endVisit} methods for each node type, that is
2310 called after the \method{visit} method for a node. In addition to these visit
2311 methods, there are also the methods \method{preVisit(ASTNode)},
2312 \method{postVisit(ASTNode)} and \method{preVisit2(ASTNode)}. The
2313 \method{preVisit} method is called before the type-specific \method{visit}
2314 method. The \method{postVisit} method is called after the type-specific
2315 \method{endVisit}. The type specific \method{visit} is only called if
2316 \method{preVisit2} returns \var{true}. Overriding the \method{preVisit2} is also
2317 altering the behavior of \method{preVisit}, since the default implementation is
2318 responsible for calling it.
2320 An example of a trivial \type{ASTVisitor} is shown in
2321 \myref{lst:astVisitorExample}.
2324 \begin{minted}{java}
2325 public class CollectNamesVisitor extends ASTVisitor {
2326 Collection<Name> names = new LinkedList<Name>();
2329 public boolean visit(QualifiedName node) {
2335 public boolean visit(SimpleName node) {
2341 \caption{An \type{ASTVisitor} that visits all the names in a subtree and adds
2342 them to a collection, except those names that are children of any
2343 \type{QualifiedName}.}
2344 \label{lst:astVisitorExample}
2347 \section{Property collectors}\label{propertyCollectors}
2348 The prefixes and unfixes are found by property
2349 collectors\typeref{no.uio.ifi.refaktor.extractors.collectors.PropertyCollector}.
2350 A property collector is of the \type{ASTVisitor} type, and thus visits nodes of
2351 type \type{ASTNode} of the abstract syntax tree \see{astVisitor}.
2353 \subsection{The PrefixesCollector}
2354 The \typewithref{no.uio.ifi.refaktor.extractors.collectors}{PrefixesCollector}
2355 finds prefixes that makes up the basis for calculating move targets for the
2356 Extract and Move Method refactoring. It visits expression
2357 statements\typeref{org.eclipse.jdt.core.dom.ExpressionStatement} and creates
2358 prefixes from its expressions in the case of method invocations. The prefixes
2359 found is registered with a prefix set, together with all its sub-prefixes.
2361 \subsection{The UnfixesCollector}\label{unfixes}
2362 The \typewithref{no.uio.ifi.refaktor.extractors.collectors}{UnfixesCollector}
2363 finds unfixes within a selection. That is prefixes that cannot be used as a
2364 basis for finding a move target in a refactoring.
2366 An unfix can be a name that is assigned to within a selection. The reason that
2367 this cannot be allowed, is that the result would be an assignment to the
2368 \type{this} keyword, which is not valid in Java \see{eclipse_bug_420726}.
2370 Prefixes that originates from variable declarations within the same selection
2371 are also considered unfixes. This is because when a method is moved, it needs to
2372 be called through a variable. If this variable is also within the method that is
2373 to be moved, this obviously cannot be done.
2375 Also considered as unfixes are variable references that are of types that is not
2376 suitable for moving a methods to. This can be either because it is not
2377 physically possible to move the method to the desired class or that it will
2378 cause compilation errors by doing so.
2380 If the type binding for a name is not resolved it is considered and unfix. The
2381 same applies to types that is only found in compiled code, so they have no
2382 underlying source that is accessible to us. (E.g. the \type{java.lang.String}
2385 Interfaces types are not suitable as targets. This is simply because interfaces
2386 in java cannot contain methods with bodies. (This thesis does not deal with
2387 features of Java versions later than Java 7. Java 8 has interfaces with default
2388 implementations of methods.) Neither are local types allowed. This accounts for
2389 both local and anonymous classes. Anonymous classes are effectively the same as
2390 interface types with respect to unfixes. Local classes could in theory be used
2391 as targets, but this is not possible due to limitations of the implementation of
2392 the Extract and Move Method refactoring. The problem is that the refactoring is
2393 done in two steps, so the intermediate state between the two refactorings would
2394 not be legal Java code. In the case of local classes, the problem is that, in
2395 the intermediate step, a selection referencing a local class would need to take
2396 the local class as a parameter if it were to be extracted to a new method. This
2397 new method would need to live in the scope of the declaring class of the
2398 originating method. The local class would then not be in the scope of the
2399 extracted method, thus bringing the source code into an illegal state. One could
2400 imagine that the method was extracted and moved in one operation, without an
2401 intermediate state. Then it would make sense to include variables with types of
2402 local classes in the set of legal targets, since the local classes would then be
2403 in the scopes of the method calls. If this makes any difference for software
2404 metrics that measure coupling would be a different discussion.
2407 \begin{multicols}{2}
2408 \begin{minted}[]{java}
2410 void declaresLocalClass() {
2425 \begin{minted}[]{java}
2426 // After Extract Method
2427 void declaresLocalClass() {
2438 // Intermediate step
2439 void fooBar(LocalClass inst) {
2445 \caption{When Extract and Move Method tries to use a variable with a local type
2446 as the move target, an intermediate step is taken that is not allowed. Here:
2447 \type{LocalClass} is not in the scope of \method{fooBar} in its intermediate
2449 \label{lst:extractMethod_LocalClass}
2452 The last class of names that are considered unfixes is names used in null tests.
2453 These are tests that reads like this: if \texttt{<name>} equals \var{null} then
2454 do something. If allowing variables used in those kinds of expressions as
2455 targets for moving methods, we would end up with code containing boolean
2456 expressions like \texttt{this == null}, which would not be meaningful, since
2457 \var{this} would never be \var{null}.
2460 \subsection{The ContainsReturnStatementCollector}
2462 \typewithref{no.uio.ifi.refaktor.analyze.collectors}{ContainsReturnStatementCollector}
2463 is a very simple property collector. It only visits the return statements within
2464 a selection, and can report whether it encountered a return statement or not.
2466 \subsection{The LastStatementCollector}
2467 The \typewithref{no.uio.ifi.refaktor.analyze.collectors}{LastStatementCollector}
2468 collects the last statement of a selection. It does so by only visiting the top
2469 level statements of the selection, and compares the textual end offset of each
2470 encuntered statement with the end offset of the previous statement found.
2472 \section{Checkers}\label{checkers}
2473 The checkers are a range of classes that checks that selections complies with
2474 certian criterias. If a
2475 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{Checker} fails, it throws a
2476 \type{CheckerException}. The checkers are managed by the
2477 \type{LegalStatementsChecker}, which does not, in fact, implement the
2478 \type{Checker} interface. It does, however, run all the checkers registered with
2479 it, and reports that all statements are considered legal if no
2480 \type{CheckerException} is thrown. Many of the checkers either extends the
2481 \type{PropertyCollector} or utilizes one or more property collectors to verify
2482 some criterias. The checkers registered with the \type{LegalStatementsChecker}
2483 are described next. They are run in the order presented below.
2485 \subsection{The EnclosingInstanceReferenceChecker}
2486 The purpose of this checker is to verify that the names in a selection is not
2487 referencing any enclosing instances. This is for making sure that all references
2488 is legal in a method that is to be moved. Theoretically, some situations could
2489 be easily solved my passing a reference to the referenced class with the moved
2490 method (e.g. when calling public methods), but the dependency on the
2491 \type{MoveInstanceMethodProcessor} prevents this.
2494 \typewithref{no.uio.ifi.refaktor.analyze.analyzers}{EnclosingInstanceReferenceChecker}
2495 is a modified version of the
2496 \typewithref{org.eclipse.jdt.internal.corext.refactoring.structure.MoveInstanceMethodProcessor}{EnclosingInstanceReferenceFinder}
2497 from the \type{MoveInstanceMethodProcessor}. Wherever the
2498 \type{EnclosingInstanceReferenceFinder} would create a fatal error status, the
2499 checker throws a \type{CheckerException}.
2501 It works by first finding all of the enclosing types of a selection. Thereafter
2502 it visits all its simple names to check that they are not references to
2503 variables or methods declared in any of the enclosing types. In addition the
2504 checker visits \var{this}-expressions to verify that no such expressions is
2505 qualified with any name.
2507 \subsection{The ReturnStatementsChecker}\label{returnStatementsChecker}
2508 \todoin{Write\ldots/change implementation/use control flow graph?}
2510 \subsection{The AmbiguousReturnValueChecker}
2511 This checker verifies that there are no \emph{ambiguous return statements} in a
2512 selection. The problem with ambiguous return statements arise when a selection
2513 is chosen to be extracted into a new method, but it needs to return more than
2514 one value from that method. This problem occurs in two situations. The first
2515 situation arise when there is more than one local variable that is both assigned
2516 to within a selection and also referenced after the selection. The other
2517 situation occur when there is only one such assignment, but there is also one or
2518 more return statements in the selection.
2520 First the checker needs to collect some data. Those data are the binding keys
2521 for all simple names that are assigned to within the selection, including
2522 variable declarations, but excluding fields. The checker also collects whether
2523 there exists a return statement in the selection or not. No further checks of
2524 return statements are needed, since, at this point, the selection is already
2525 checked for illegal return statements \see{returnStatementsChecker}.
2527 After the binding keys of the assignees are collected, the checker searches the
2528 part of the enclosing method that is after the selection for references whose
2529 binding keys are among the the collected keys. If more than one unique referral
2530 is found, or only one referral is found, but the selection also contains a
2531 return statement, we have a situation with an ambiguous return value, and an
2532 exception is thrown.
2534 %\todoin{Explain why we do not need to consider variables assigned inside
2535 %local/anonymous classes. (The referenced variables need to be final and so
2538 \subsection{The IllegalStatementsChecker}
2539 This checker is designed to check for illegal statements.
2541 Any use of the \var{super} keyword is prohibited, since its meaning is altered
2542 when moving a method to another class.
2544 For a \emph{break} statement, there is two situations to consider: A break
2545 statement with or without a label. If the break statement has a label, it is
2546 checked that whole of the labeled statement is inside the selection. Since a
2547 label does not have any binding information, we have to search upwards in the
2548 AST to find the \type{LabeledStatement} that corresponds to the label from the
2549 break statement, and check that it is contained in the selection. If the break
2550 statement does not have a label attached to it, it is checked that its innermost
2551 enclosing loop or switch statement also is inside the selection.
2553 The situation for a \emph{continue} statement is the same as for a break
2554 statement, except that it is not allowed inside switch statements.
2556 Regarding \emph{assignments}, two types of assignments is allowed: Assignment to
2557 a non-final variable and assignment to an array access. All other assignments is
2560 \todoin{Finish\ldots}
2563 \chapter{Benchmarking}
2564 \todoin{Better name than ``benchmarking''?}
2565 This part of the master project is located in the Eclipse project
2566 \code{no.uio.ifi.refaktor.benchmark}. The purpose of it is to run the equivalent
2567 of the \type{SearchBasedExtractAndMoveMethodChanger}
2568 \see{searchBasedExtractAndMoveMethodChanger} over a larger software project,
2569 both to test its roubustness but also its effect on different software metrics.
2571 \section{The benchmark setup}
2572 The benchmark itself is set up as a \emph{JUnit} test case. This is a convenient
2573 setup, and utilizes the \emph{JUnit Plugin Test Launcher}. This provides us a
2574 with a fully functional Eclipse workbench. Most importantly, this gives us
2575 access to the Java Model of Eclipse \see{javaModel}.
2577 \subsection{The ProjectImporter}
2578 The Java project that is going to be used as the data for the benchmark, must be
2579 imported into the JUnit workspace. This is done by the
2580 \typewithref{no.uio.ifi.refaktor.benchmark}{ProjectImporter}. The importer
2581 require the absolute path to the project description file. It is named
2582 \code{.project} and is located at the root of the project directory.
2584 The project description is loaded to find the name of the project to be
2585 imported. The project that shall be the destination for the import is created in
2586 the workspace, on the base of the name from the description. Then an import
2587 operation is created, based on both the source and destination information. The
2588 import operation is run to perform the import.
2590 I have found no simple API call to accomplish what the importer does, which
2591 tells me that it may not be too many people performing this particular action.
2592 The solution to the problem was found on \emph{Stack
2593 Overflow}\footnote{\url{https://stackoverflow.com/questions/12401297}}. It
2594 contains enough dirty details to be considered unconvenient to use, if not
2595 wrapping it in a class like my \type{ProjectImporter}. One would probably have
2596 to delve into the source code for the import wizard to find out how the import
2597 operation works, if no one had already done it.
2599 \section{Statistics}
2600 Statistics for the analysis and changes is captured by the
2601 \typewithref{no.uio.ifi.refaktor.aspects}{StatisticsAspect}. This an
2602 \emph{aspect} written in \emph{AspectJ}.
2604 \subsection{AspectJ}
2605 \emph{AspectJ}\footnote{\url{http://eclipse.org/aspectj/}} is an extension to
2606 the Java language, and facilitates combining aspect-oriented programming with
2607 the object-oriented programming in Java.
2609 Aspect-oriented programming is a programming paradigm that is meant to isolate
2610 so-called \emph{cross-cutting concerns} into their own modules. These
2611 cross-cutting concerns are functionalities that spans over multiple classes, but
2612 may not belong naturally in any of them. It can be functionality that does not
2613 concern the business logic of an application, and thus may be a burden when
2614 entangled with parts of the source code it does not really belong. Examples
2615 include logging, debugging, optimization and security.
2617 Aspects are interacting with other modules by defining advices. The concept of
2618 an \emph{advice} is known from both aspect-oriented and functional
2619 programming\citing{wikiAdvice2014}. It is a function that modifies another
2620 function when the latter is run. An advice in AspectJ is somewhat similar to a
2621 method in Java. It is meant to alter the behavior of other methods, and contains
2622 a body that is executed when it is applied.
2624 An advice can be applied at a defined \emph{pointcut}. A pointcut picks out one
2625 or more \emph{join points}. A join point is a well-defined point in the
2626 execution of a program. It can occur when calling a method defined for a
2627 particular class, when calling all methods with the same name,
2628 accessing/assigning to a particular field of a given class and so on. An advice
2629 can be declared to run both before, after returning from a pointcut, when there
2630 is thrown an exception in the pointcut or after the pointcut either returns or
2631 throws an exception. In addition to picking out join points, a pointcut can
2632 also bind variables from its context, so they can be accessed in the body of an
2633 advice. An example of a pointcut and an advice is found in
2634 \myref{lst:aspectjExample}.
2637 \begin{minted}{aspectj}
2638 pointcut methodAnalyze(
2639 SearchBasedExtractAndMoveMethodAnalyzer analyzer) :
2640 call(* SearchBasedExtractAndMoveMethodAnalyzer.analyze())
2641 && target(analyzer);
2643 after(SearchBasedExtractAndMoveMethodAnalyzer analyzer) :
2644 methodAnalyze(analyzer) {
2645 statistics.methodCount++;
2646 debugPrintMethodAnalysisProgress(analyzer.method);
2649 \caption{An example of a pointcut named \method{methodAnalyze},
2650 and an advice defined to be applied after it has occurred.}
2651 \label{lst:aspectjExample}
2654 \subsection{The Statistics class}
2655 The statistics aspect stores statistical information in an object of type
2656 \type{Statistics}. As of now, the aspect needs to be initialized at the point in
2657 time where it is desired that it starts its data gathering. At any point in time
2658 the statistics aspect can be queried for a snapshot of the current statistics.
2660 The \type{Statistics} class also include functionality for generating a report
2661 of its gathered statistics. The report can be given either as a string or it can
2662 be written to a file.
2664 \subsection{Advices}
2665 The statistics aspect contains advices for gathering statistical data from
2666 different parts of the benchmarking process. It captures statistics from both
2667 the analysis part and the execution part of the composite \ExtractAndMoveMethod
2670 For the analysis part, there are advices to count the number of text selections
2671 analyzed and the number of methods, types, compilation units and packages
2672 analyzed. There are also advices that counts for how many of the methods there
2673 is found a selection that is a candidate for the refactoring, and for how many
2674 ethods there is not.
2676 There exists advices for counting both the successful and unsuccessful
2677 executions of all the refactorings. Both for the \ExtractMethod and \MoveMethod
2678 refactorings in isolation, as well as for the combination of them.
2680 \section{Optimizations}
2681 When looking for optimizations to make for the benchmarking process, I used the
2682 \emph{VisualVM}\footnote{\url{http://visualvm.java.net/}} for the Java Virtual
2683 Machine to both profile the application and also to make memory dumps of its
2686 \subsection{Caching}
2687 When profiling the benchmark process before making any optimizations, it early
2688 became apparent that the parsing of source code was a place to direct attention
2689 towards. This discovery was done when only \emph{analyzing} source code, before
2690 trying to do any \emph{manipulation} of it. Caching of the parsed ASTs seemed
2691 like the best way to save some time, as expected. With only a simple cache of
2692 the most recently used AST, the analysis time was speeded up by a factor of
2694 20. This number depends a little upon which type of system the analysis was
2697 The caching is managed by a cache manager, that now, by default, utilizes the
2698 not so well known feature of Java called a \emph{soft reference}. Soft
2699 references are best explained in the context of weak references. A \emph{weak
2700 reference} is a reference to an object instance that is only guaranteed to
2701 persist as long as there is a \emph{strong reference} or a soft reference
2702 referring the same object. If no such reference is found, its referred object is
2703 garbage collected. A strong reference is basically the same as a regular Java
2704 reference. A soft reference has the same guarantees as a week reference when it
2705 comes to its relation to strong references, but it is not necessarily garbage
2706 collected whenever there exists no strong references to it. A soft reference
2707 \emph{may} reside in memory as long as the JVM has enough free memory in the
2708 heap. A soft reference will therefore usually perform better than a weak
2709 reference when used for simple caching and similar tasks. The way to use a
2710 soft/weak reference is to as it for its referent. The return value then has to
2711 be tested to check that it is not \var{null}. For the basic usage of soft
2712 references, see \myref{lst:softReferenceExample}. For a more thorough
2713 explanation of weak references in general, see\citing{weakRef2006}.
2716 \begin{minted}{java}
2718 Object strongRef = new Object();
2721 SoftReference<Object> softRef =
2722 new SoftReference<Object>(new Object());
2724 // Using the soft reference
2725 Object obj = softRef.get();
2730 \caption{Showing the basic usage of soft references. Weak references is used the
2731 same way. {\footnotesize (The references are part of the \code{java.lang.ref}
2733 \label{lst:softReferenceExample}
2736 The cache based on soft references has no limit for how many ASTs it caches. It
2737 is generally not advisable to keep references to ASTs for prolonged periods of
2738 time, since they are expensive structures to hold on to. For regular plugin
2739 development, Eclipse recommends not creating more than one AST at a time to
2740 limit memory consumption. Since the benchmarking has nothing to do with user
2741 experience, and throughput is everything, these advices are intentionally
2742 ignored. This means that during the benchmarking process, the target Eclipse
2743 application may very well work close to its memory limit for the heap space for
2744 long periods during the benchmark.
2746 \subsection{Memento}
2748 \chapter{Eclipse Bugs Found}
2749 \todoin{Add other things and change headline?}
2751 \section{Eclipse bug 420726: Code is broken when moving a method that is
2752 assigning to the parameter that is also the move
2753 destination}\label{eclipse_bug_420726}
2754 This bug\footnote{\url{https://bugs.eclipse.org/bugs/show\_bug.cgi?id=420726}}
2755 was found when analyzing what kinds of names that was to be considered as
2756 \emph{unfixes} \see{unfixes}.
2758 \subsection{The bug}
2759 The bug emerges when trying to move a method from one class to another, and when
2760 the target for the move (must be a variable, local or field) is both a parameter
2761 variable and also is assigned to within the method body. Eclipse allows this to
2762 happen, although it is the sure path to a compilation error. This is because we
2763 would then have an assignment to a \var{this} expression, which is not allowed
2766 \subsection{The solution}
2767 The solution to this problem is to add all simple names that are assigned to in
2768 a method body to the set of unfixes.
2770 \section{Eclipse bug 429416: IAE when moving method from anonymous class}
2772 discovered\footnote{\url{https://bugs.eclipse.org/bugs/show\_bug.cgi?id=429416}}
2773 this bug during a batch change on the \type{org.eclipse.jdt.ui} project.
2775 \subsection{The bug}
2776 This bug surfaces when trying to use the Move Method refactoring to move a
2777 method from an anonymous class to another class. This happens both for my
2778 simulation as well as in Eclipse, through the user interface. It only occurs
2779 when Eclipse analyzes the program and finds it necessary to pass an instance of
2780 the originating class as a parameter to the moved method. I.e. it want to pass a
2781 \var{this} expression. The execution ends in an
2782 \typewithref{java.lang}{IllegalArgumentException} in
2783 \typewithref{org.eclipse.jdt.core.dom}{SimpleName} and its
2784 \method{setIdentifier(String)} method. The simple name is attempted created in
2786 \methodwithref{org.eclipse.jdt.internal.corext.refactoring.structure.\\MoveInstanceMethodProcessor}{createInlinedMethodInvocation}
2787 so the \type{MoveInstanceMethodProcessor} was early a clear suspect.
2789 The \method{createInlinedMethodInvocation} is the method that creates a method
2790 invocation where the previous invocation to the method that was moved was. From
2791 its code it can be read that when a \var{this} expression is going to be passed
2792 in to the invocation, it shall be qualified with the name of the original
2793 method's declaring class, if the declaring class is either an anonymous clas or
2794 a member class. The problem with this, is that an anonymous class does not have
2795 a name, hence the term \emph{anonymous} class! Therefore, when its name, an
2796 empty string, is passed into
2797 \methodwithref{org.eclipse.jdt.core.dom.AST}{newSimpleName} it all ends in an
2798 \type{IllegalArgumentException}.
2800 \subsection{How I solved the problem}
2801 Since the \type{MoveInstanceMethodProcessor} is instantiated in the
2802 \typewithref{no.uio.ifi.refaktor.change.executors}{MoveMethod\-RefactoringExecutor},
2803 and only need to be a
2804 \typewithref{org.eclipse.ltk.core.refactoring.participants}{MoveProcessor}, I
2805 was able to copy the code for the original move processor and modify it so that
2806 it works better for me. It is now called
2807 \typewithref{no.uio.ifi.refaktor.refactorings.processors}{ModifiedMoveInstanceMethodProcessor}.
2808 The only modification done (in addition to some imports and suppression of
2809 warnings), is in the \method{createInlinedMethodInvocation}. When the declaring
2810 class of the method to move is anonymous, the \var{this} expression in the
2811 parameter list is not qualified with the declaring class' (empty) name.
2813 \section{Eclipse bug 429954: Extracting statement with reference to local type
2814 breaks code}\label{eclipse_bug_429954}
2815 The bug\footnote{\url{https://bugs.eclipse.org/bugs/show\_bug.cgi?id=429954}}
2816 was discovered when doing some changes to the way unfixes is computed.
2818 \subsection{The bug}
2819 The problem is that Eclipse is allowing selections that references variables of
2820 local types to be extracted. When this happens the code is broken, since the
2821 extracted method must take a parameter of a local type that is not in the
2822 methods scope. The problem is illustrated in
2823 \myref{lst:extractMethod_LocalClass}, but there in another setting.
2825 \subsection{Actions taken}
2826 There are no actions directly springing out of this bug, since the Extract
2827 Method refactoring cannot be meant to be this way. This is handled on the
2828 analysis stage of our Extract and Move Method refactoring. So names representing
2829 variables of local types is considered unfixes \see{unfixes}.
2830 \todoin{write more when fixing this in legal statements checker}
2832 \chapter{Related Work}
2834 \section{The compositional paradigm of refactoring}
2835 This paradigm builds upon the observation of Vakilian et
2836 al.\citing{vakilian2012}, that of the many automated refactorings existing in
2837 modern IDEs, the simplest ones are dominating the usage statistics. The report
2838 mainly focuses on \emph{Eclipse} as the tool under investigation.
2840 The paradigm is described almost as the opposite of automated composition of
2841 refactorings \see{compositeRefactorings}. It works by providing the programmer
2842 with easily accessible primitive refactorings. These refactorings shall be
2843 accessed via keyboard shortcuts or quick-assist menus\footnote{Think
2844 quick-assist with Ctrl+1 in Eclipse} and be promptly executed, opposed to in the
2845 currently dominating wizard-based refactoring paradigm. They are ment to
2846 stimulate composing smaller refactorings into more complex changes, rather than
2847 doing a large upfront configuration of a wizard-based refactoring, before
2848 previewing and executing it. The compositional paradigm of refactoring is
2849 supposed to give control back to the programmer, by supporting \himher with an
2850 option of performing small rapid changes instead of large changes with a lesser
2851 degree of control. The report authors hope this will lead to fewer unsuccessful
2852 refactorings. It also could lower the bar for understanding the steps of a
2853 larger composite refactoring and thus also help in figuring out what goes wrong
2854 if one should choose to op in on a wizard-based refactoring.
2856 Vakilian and his associates have performed a survey of the effectiveness of the
2857 compositional paradigm versus the wizard-based one. They claim to have found
2858 evidence of that the \emph{compositional paradigm} outperforms the
2859 \emph{wizard-based}. It does so by reducing automation, which seem
2860 counterintuitive. Therefore they ask the question ``What is an appropriate level
2861 of automation?'', and thus questions what they feel is a rush toward more
2862 automation in the software engineering community.