8 This plan is *provisional*. It is not agreed upon. It is written with the
9 intention of capturing the desires and concerns of the LLVM community, and
10 forming them into a plan that can be agreed upon.
11 The original author is somewhat naïve in the ways of LLVM so there will
12 inevitably be some details that are flawed. You can help - you can edit this
13 page (preferably with a Phabricator review for larger changes) or reply to the
14 `Request For Comments thread
15 <http://lists.llvm.org/pipermail/llvm-dev/2019-February/130083.html>`_.
20 Improve the readability of LLVM code.
25 The current `variable naming rule
26 <../CodingStandards.html#name-types-functions-variables-and-enumerators-properly>`_
29 Variable names should be nouns (as they represent state). The name should be
30 camel case, and start with an upper case letter (e.g. Leader or Boats).
32 This rule is the same as that for type names. This is a problem because the
33 type name cannot be reused for a variable name [*]_. LLVM developers tend to
34 work around this by either prepending ``The`` to the type name::
38 ... or more commonly use an acronym, despite the coding standard stating "Avoid
39 abbreviations unless they are well known"::
43 The proliferation of acronyms leads to hard-to-read code such as `this
44 <https://github.com/llvm/llvm-project/blob/0a8bc14ad7f3209fe702d18e250194cd90188596/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp#L7445>`_::
46 InnerLoopVectorizer LB(L, PSE, LI, DT, TLI, TTI, AC, ORE, VF.Width, IC,
49 Many other coding guidelines [LLDB]_ [Google]_ [WebKit]_ [Qt]_ [Rust]_ [Swift]_
50 [Python]_ require that variable names begin with a lower case letter in contrast
51 to class names which begin with a capital letter. This convention means that the
52 most readable variable name also requires the least thought::
56 There is some agreement that the current rule is broken [LattnerAgree]_
57 [ArsenaultAgree]_ [RobinsonAgree]_ and that acronyms are an obstacle to reading
58 new code [MalyutinDistinguish]_ [CarruthAcronym]_ [PicusAcronym]_. There are
59 some opposing views [ParzyszekAcronym2]_ [RicciAcronyms]_.
61 This work-in-progress proposal is to change the coding standard for variable
62 names to require that they start with a lower case letter.
65 <https://github.com/llvm/llvm-project/blob/8b72080d4d7b13072f371712eed333f987b7a18e/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp#L2727>`_
66 the type name *is* reused as a variable name, but this shadows the type name
67 and confuses many debuggers [DenisovCamelBack]_.
69 Variable Names Coding Standard Options
70 ======================================
72 There are two main options for variable names that begin with a lower case
73 letter: ``camelBack`` and ``lower_case``. (These are also known by other names
74 but here we use the terminology from clang-tidy).
76 ``camelBack`` is consistent with [WebKit]_, [Qt]_ and [Swift]_ while
77 ``lower_case`` is consistent with [LLDB]_, [Google]_, [Rust]_ and [Python]_.
79 ``camelBack`` is already used for function names, which may be considered an
80 advantage [LattnerFunction]_ or a disadvantage [CarruthFunction]_.
82 Approval for ``camelBack`` was expressed by [DenisovCamelBack]_
83 [LattnerFunction]_ [IvanovicDistinguish]_.
84 Opposition to ``camelBack`` was expressed by [CarruthCamelBack]_
86 Approval for ``lower_case`` was expressed by [CarruthLower]_
87 [CarruthCamelBack]_ [TurnerLLDB]_.
88 Opposition to ``lower_case`` was expressed by [LattnerLower]_.
90 Differentiating variable kinds
91 ------------------------------
93 An additional requested change is to distinguish between different kinds of
94 variables [RobinsonDistinguish]_ [RobinsonDistinguish2]_ [JonesDistinguish]_
95 [IvanovicDistinguish]_ [CarruthDistinguish]_ [MalyutinDistinguish]_.
97 Others oppose this idea [HähnleDistinguish]_ [GreeneDistinguish]_
100 A possibility is for member variables to be prefixed with ``m_`` and for global
101 variables to be prefixed with ``g_`` to distinguish them from local variables.
102 This is consistent with [LLDB]_. The ``m_`` prefix is consistent with [WebKit]_.
104 A variation is for member variables to be prefixed with ``m``
105 [IvanovicDistinguish]_ [BeylsDistinguish]_. This is consistent with [Mozilla]_.
107 Another option is for member variables to be suffixed with ``_`` which is
108 consistent with [Google]_ and similar to [Python]_. Opposed by
109 [ParzyszekDistinguish]_.
111 Reducing the number of acronyms
112 ===============================
114 While switching coding standard will make it easier to use non-acronym names for
115 new code, it doesn't improve the existing large body of code that uses acronyms
116 extensively to the detriment of its readability. Further, it is natural and
117 generally encouraged that new code be written in the style of the surrounding
118 code. Therefore it is likely that much newly written code will also use
119 acronyms despite what the coding standard says, much as it is today.
121 As well as changing the case of variable names, they could also be expanded to
122 their non-acronym form e.g. ``Triple T`` → ``Triple triple``.
124 There is support for expanding many acronyms [CarruthAcronym]_ [PicusAcronym]_
125 but there is a preference that expanding acronyms be deferred
126 [ParzyszekAcronym]_ [CarruthAcronym]_.
128 The consensus within the community seems to be that at least some acronyms are
129 valuable [ParzyszekAcronym]_ [LattnerAcronym]_. The most commonly cited acronym
130 is ``TLI`` however that is used to refer to both ``TargetLowering`` and
131 ``TargetLibraryInfo`` [GreeneDistinguish]_.
133 The following is a list of acronyms considered sufficiently useful that the
134 benefit of using them outweighs the cost of learning them. Acronyms that are
135 either not on the list or are used to refer to a different type should be
138 ============================ =============
139 Class name Variable name
140 ============================ =============
141 DeterministicFiniteAutomaton dfa
146 MachineRegisterInfo mri
149 TargetLibraryInfo tli
150 TargetRegisterInfo tri
151 ============================ =============
153 In some cases renaming acronyms to the full type name will result in overly
154 verbose code. Unlike most classes, a variable's scope is limited and therefore
155 some of its purpose can implied from that scope, meaning that fewer words are
156 necessary to give it a clear name. For example, in an optization pass the reader
157 can assume that a variable's purpose relates to optimization and therefore an
158 ``OptimizationRemarkEmitter`` variable could be given the name ``remarkEmitter``
159 or even ``remarker``.
161 The following is a list of longer class names and the associated shorter
164 ========================= =============
165 Class name Variable name
166 ========================= =============
169 ExecutionEngine engine
170 MachineOperand operand
171 OptimizationRemarkEmitter remarker
172 PreservedAnalyses analyses
173 PreservedAnalysesChecker checker
174 TargetLowering lowering
175 TargetMachine machine
176 ========================= =============
181 There are three main options for transitioning:
183 1. Keep the current coding standard
187 Keep the current coding standard
188 --------------------------------
190 Proponents of keeping the current coding standard (i.e. not transitioning at
191 all) question whether the cost of transition outweighs the benefit
192 [EmersonConcern]_ [ReamesConcern]_ [BradburyConcern]_.
193 The costs are that ``git blame`` will become less usable; and that merging the
194 changes will be costly for downstream maintainers. See `Big bang`_ for potential
200 The coding standard could allow both ``CamelCase`` and ``camelBack`` styles for
201 variable names [LattnerTransition]_.
203 A code review to implement this is at https://reviews.llvm.org/D57896.
208 * Very easy to implement initially.
213 * Leads to inconsistency [BradburyConcern]_ [AminiInconsistent]_.
214 * Inconsistency means it will be hard to know at a guess what name a variable
215 will have [DasInconsistent]_ [CarruthInconsistent]_.
216 * Some large-scale renaming may happen anyway, leading to its disadvantages
217 without any mitigations.
222 With this approach, variables will be renamed by an automated script in a series
225 The principle advantage of this approach is that it minimises the cost of
226 inconsistency [BradburyTransition]_ [RobinsonTransition]_.
228 It goes against a policy of avoiding large-scale reformatting of existing code
229 [GreeneDistinguish]_.
231 It has been suggested that LLD would be a good starter project for the renaming
234 Keeping git blame usable
235 ************************
237 ``git blame`` (or ``git annotate``) permits quickly identifying the commit that
238 changed a given line in a file. After renaming variables, many lines will show
239 as being changed by that one commit, requiring a further invocation of ``git
240 blame`` to identify prior, more interesting commits [GreeneGitBlame]_
243 **Mitigation**: `git-hyper-blame
244 <https://commondatastorage.googleapis.com/chrome-infra-docs/flat/depot_tools/docs/html/git-hyper-blame.html>`_
245 can ignore or "look through" a given set of commits.
246 A ``.git-blame-ignore-revs`` file identifying the variable renaming commits
247 could be added to the LLVM git repository root directory.
248 It is being `investigated
249 <https://public-inbox.org/git/20190324235020.49706-1-michael@platin.gs/>`_
250 whether similar functionality could be added to ``git blame`` itself.
252 Minimising cost of downstream merges
253 ************************************
255 There are many forks of LLVM with downstream changes. Merging a large-scale
256 renaming change could be difficult for the fork maintainers.
258 **Mitigation**: A large-scale renaming would be automated. A fork maintainer can
259 merge from the commit immediately before the renaming, then apply the renaming
260 script to their own branch. They can then merge again from the renaming commit,
261 resolving all conflicts by choosing their own version. This could be tested on
267 This is a provisional plan for the `Big bang`_ approach. It has not been agreed.
269 #. Investigate improving ``git blame``. The extent to which it can be made to
270 "look through" commits may impact how big a change can be made.
272 #. Write a script to expand acronyms.
274 #. Experiment and perform dry runs of the various refactoring options.
275 Results can be published in forks of the LLVM Git repository.
277 #. Consider the evidence and agree on the new policy.
279 #. Agree & announce a date for the renaming of the starter project (LLD).
281 #. Update the `policy page <../CodingStandards.html>`_. This will explain the
282 old and new rules and which projects each applies to.
284 #. Refactor the starter project in two commits:
286 1. Add or change the project's .clang-tidy to reflect the agreed rules.
287 (This is in a separate commit to enable the merging process described in
288 `Minimising cost of downstream merges`_).
289 Also update the project list on the policy page.
290 2. Apply ``clang-tidy`` to the project's files, with only the
291 ``readability-identifier-naming`` rules enabled. ``clang-tidy`` will also
292 reformat the affected lines according to the rules in ``.clang-format``.
293 It is anticipated that this will be a good dog-fooding opportunity for
294 clang-tidy, and bugs should be fixed in the process, likely including:
296 * `readability-identifier-naming incorrectly fixes lambda capture
297 <https://bugs.llvm.org/show_bug.cgi?id=41119>`_.
298 * `readability-identifier-naming incorrectly fixes variables which
299 become keywords <https://bugs.llvm.org/show_bug.cgi?id=41120>`_.
300 * `readability-identifier-naming misses fixing member variables in
301 destructor <https://bugs.llvm.org/show_bug.cgi?id=41122>`_.
303 #. Gather feedback and refine the process as appropriate.
305 #. Apply the process to the following projects, with a suitable delay between
306 each (at least 4 weeks after the first change, at least 2 weeks subsequently)
307 to allow gathering further feedback.
308 This list should exclude projects that must adhere to an externally defined
309 standard e.g. libcxx.
310 The list is roughly in chronological order of renaming.
311 Some items may not make sense to rename individually - it is expected that
312 this list will change following experimentation:
332 * WebAssembly backend
345 #. Remove the old variable name rule from the policy page.
347 #. Repeat many of the steps in the sequence, using a script to expand acronyms.
352 .. [LLDB] LLDB Coding Conventions https://llvm.org/svn/llvm-project/lldb/branches/release_39/www/lldb-coding-conventions.html
353 .. [Google] Google C++ Style Guide https://google.github.io/styleguide/cppguide.html#Variable_Names
354 .. [WebKit] WebKit Code Style Guidelines https://webkit.org/code-style-guidelines/#names
355 .. [Qt] Qt Coding Style https://wiki.qt.io/Qt_Coding_Style#Declaring_variables
356 .. [Rust] Rust naming conventions https://doc.rust-lang.org/1.0.0/style/style/naming/README.html
357 .. [Swift] Swift API Design Guidelines https://swift.org/documentation/api-design-guidelines/#general-conventions
358 .. [Python] Style Guide for Python Code https://www.python.org/dev/peps/pep-0008/#function-and-variable-names
359 .. [Mozilla] Mozilla Coding style: Prefixes https://developer.mozilla.org/en-US/docs/Mozilla/Developer_guide/Coding_Style#Prefixes
360 .. [SVE] LLVM with support for SVE https://github.com/ARM-software/LLVM-SVE
361 .. [AminiInconsistent] Mehdi Amini, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130329.html
362 .. [ArsenaultAgree] Matt Arsenault, http://lists.llvm.org/pipermail/llvm-dev/2019-February/129934.html
363 .. [BeylsDistinguish] Kristof Beyls, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130292.html
364 .. [BradburyConcern] Alex Bradbury, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130266.html
365 .. [BradburyTransition] Alex Bradbury, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130388.html
366 .. [CarruthAcronym] Chandler Carruth, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130313.html
367 .. [CarruthCamelBack] Chandler Carruth, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130214.html
368 .. [CarruthDistinguish] Chandler Carruth, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130310.html
369 .. [CarruthFunction] Chandler Carruth, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130309.html
370 .. [CarruthInconsistent] Chandler Carruth, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130312.html
371 .. [CarruthLower] Chandler Carruth, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130430.html
372 .. [DasInconsistent] Sanjoy Das, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130304.html
373 .. [DenisovCamelBack] Alex Denisov, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130179.html
374 .. [EmersonConcern] Amara Emerson, http://lists.llvm.org/pipermail/llvm-dev/2019-February/129894.html
375 .. [GreeneDistinguish] David Greene, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130425.html
376 .. [GreeneGitBlame] David Greene, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130228.html
377 .. [HendersonPrefix] James Henderson, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130465.html
378 .. [HähnleDistinguish] Nicolai Hähnle, http://lists.llvm.org/pipermail/llvm-dev/2019-February/129923.html
379 .. [IvanovicDistinguish] Nemanja Ivanovic, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130249.html
380 .. [JonesDistinguish] JD Jones, http://lists.llvm.org/pipermail/llvm-dev/2019-February/129926.html
381 .. [LattnerAcronym] Chris Lattner, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130353.html
382 .. [LattnerAgree] Chris Latter, http://lists.llvm.org/pipermail/llvm-dev/2019-February/129907.html
383 .. [LattnerFunction] Chris Lattner, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130630.html
384 .. [LattnerLower] Chris Lattner, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130629.html
385 .. [LattnerTransition] Chris Lattner, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130355.html
386 .. [MalyutinDistinguish] Danila Malyutin, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130320.html
387 .. [ParzyszekAcronym] Krzysztof Parzyszek, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130306.html
388 .. [ParzyszekAcronym2] Krzysztof Parzyszek, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130323.html
389 .. [ParzyszekDistinguish] Krzysztof Parzyszek, http://lists.llvm.org/pipermail/llvm-dev/2019-February/129941.html
390 .. [PicusAcronym] Diana Picus, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130318.html
391 .. [ReamesConcern] Philip Reames, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130181.html
392 .. [RicciAcronyms] Bruno Ricci, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130328.html
393 .. [RobinsonAgree] Paul Robinson, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130111.html
394 .. [RobinsonDistinguish] Paul Robinson, http://lists.llvm.org/pipermail/llvm-dev/2019-February/129920.html
395 .. [RobinsonDistinguish2] Paul Robinson, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130229.html
396 .. [RobinsonTransition] Paul Robinson, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130415.html
397 .. [TurnerCamelBack] Zachary Turner, https://reviews.llvm.org/D57896#1402264
398 .. [TurnerLLDB] Zachary Turner, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130213.html
399 .. [Ueyama] Rui Ueyama, http://lists.llvm.org/pipermail/llvm-dev/2019-February/130435.html