11 This document contains the LLVM Developer Policy which defines the project's
12 policy towards developers and their contributions. The intent of this policy is
13 to eliminate miscommunication, rework, and confusion that might arise from the
14 distributed nature of LLVM's development. By stating the policy in clear terms,
15 we hope each developer can know ahead of time what to expect when making LLVM
16 contributions. This policy covers all llvm.org subprojects, including Clang,
19 This policy is also designed to accomplish the following objectives:
21 #. Attract both users and developers to the LLVM project.
23 #. Make life as simple and easy for contributors as possible.
25 #. Keep the top of Subversion trees as stable as possible.
27 #. Establish awareness of the project's :ref:`copyright, license, and patent
28 policies <copyright-license-patents>` with contributors to the project.
30 This policy is aimed at frequent contributors to LLVM. People interested in
31 contributing one-off patches can do so in an informal way by sending them to the
32 `llvm-commits mailing list
33 <http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits>`_ and engaging another
34 developer to see it through the process.
39 This section contains policies that pertain to frequent LLVM developers. We
40 always welcome `one-off patches`_ from people who do not routinely contribute to
41 LLVM, but we expect more from frequent contributors to keep the system as
42 efficient as possible for everyone. Frequent LLVM contributors are expected to
43 meet the following requirements in order for LLVM to maintain a high standard of
49 Developers should stay informed by reading at least the "dev" mailing list for
50 the projects you are interested in, such as `llvmdev
51 <http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>`_ for LLVM, `cfe-dev
52 <http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev>`_ for Clang, or `lldb-dev
53 <http://lists.cs.uiuc.edu/mailman/listinfo/lldb-dev>`_ for LLDB. If you are
54 doing anything more than just casual work on LLVM, it is suggested that you also
55 subscribe to the "commits" mailing list for the subproject you're interested in,
57 <http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits>`_, `cfe-commits
58 <http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits>`_, or `lldb-commits
59 <http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits>`_. Reading the
60 "commits" list and paying attention to changes being made by others is a good
61 way to see what other people are interested in and watching the flow of the
64 We recommend that active developers register an email account with `LLVM
65 Bugzilla <http://llvm.org/bugs/>`_ and preferably subscribe to the `llvm-bugs
66 <http://lists.cs.uiuc.edu/mailman/listinfo/llvmbugs>`_ email list to keep track
67 of bugs and enhancements occurring in LLVM. We really appreciate people who are
68 proactive at catching incoming bugs in their components and dealing with them
77 When making a patch for review, the goal is to make it as easy for the reviewer
78 to read it as possible. As such, we recommend that you:
80 #. Make your patch against the Subversion trunk, not a branch, and not an old
81 version of LLVM. This makes it easy to apply the patch. For information on
82 how to check out SVN trunk, please see the `Getting Started
83 Guide <GettingStarted.html#checkout>`_.
85 #. Similarly, patches should be submitted soon after they are generated. Old
86 patches may not apply correctly if the underlying code changes between the
87 time the patch was created and the time it is applied.
89 #. Patches should be made with ``svn diff``, or similar. If you use a
90 different tool, make sure it uses the ``diff -u`` format and that it
91 doesn't contain clutter which makes it hard to read.
93 #. If you are modifying generated files, such as the top-level ``configure``
94 script, please separate out those changes into a separate patch from the rest
97 When sending a patch to a mailing list, it is a good idea to send it as an
98 *attachment* to the message, not embedded into the text of the message. This
99 ensures that your mailer will not mangle the patch when it sends it (e.g. by
100 making whitespace changes or by wrapping lines).
102 *For Thunderbird users:* Before submitting a patch, please open *Preferences >
103 Advanced > General > Config Editor*, find the key
104 ``mail.content_disposition_type``, and set its value to ``1``. Without this
105 setting, Thunderbird sends your attachment using ``Content-Disposition: inline``
106 rather than ``Content-Disposition: attachment``. Apple Mail gamely displays such
107 a file inline, making it difficult to work with for reviewers using that
115 LLVM has a code review policy. Code review is one way to increase the quality of
116 software. We generally follow these policies:
118 #. All developers are required to have significant changes reviewed before they
119 are committed to the repository.
121 #. Code reviews are conducted by email, usually on the llvm-commits list.
123 #. Code can be reviewed either before it is committed or after. We expect major
124 changes to be reviewed before being committed, but smaller changes (or
125 changes where the developer owns the component) can be reviewed after commit.
127 #. The developer responsible for a code change is also responsible for making
128 all necessary review-related changes.
130 #. Code review can be an iterative process, which continues until the patch is
131 ready to be committed. Specifically, once a patch is sent out for review, it
132 needs an explicit "looks good" before it is submitted. Do not assume silent
133 approval, or request active objections to the patch with a deadline.
135 Sometimes code reviews will take longer than you would hope for, especially for
136 larger features. Accepted ways to speed up review times for your patches are:
138 * Review other people's patches. If you help out, everybody will be more
139 willing to do the same for you; goodwill is our currency.
140 * Ping the patch. If it is urgent, provide reasons why it is important to you to
141 get this patch landed and ping it every couple of days. If it is
142 not urgent, the common courtesy ping rate is one week. Remember that you're
143 asking for valuable time from other professional developers.
144 * Ask for help on IRC. Developers on IRC will be able to either help you
145 directly, or tell you who might be a good reviewer.
146 * Split your patch into multiple smaller patches that build on each other. The
147 smaller your patch, the higher the probability that somebody will take a quick
150 Developers should participate in code reviews as both reviewers and
151 reviewees. If someone is kind enough to review your code, you should return the
152 favor for someone else. Note that anyone is welcome to review and give feedback
153 on a patch, but only people with Subversion write access can approve it.
155 There is a web based code review tool that can optionally be used
156 for code reviews. See :doc:`Phabricator`.
161 The LLVM Project relies on two features of its process to maintain rapid
162 development in addition to the high quality of its source base: the combination
163 of code review plus post-commit review for trusted maintainers. Having both is
164 a great way for the project to take advantage of the fact that most people do
165 the right thing most of the time, and only commit patches without pre-commit
166 review when they are confident they are right.
168 The trick to this is that the project has to guarantee that all patches that are
169 committed are reviewed after they go in: you don't want everyone to assume
170 someone else will review it, allowing the patch to go unreviewed. To solve this
171 problem, we have a notion of an 'owner' for a piece of the code. The sole
172 responsibility of a code owner is to ensure that a commit to their area of the
173 code is appropriately reviewed, either by themself or by someone else. The list
174 of current code owners can be found in the file
175 `CODE_OWNERS.TXT <http://llvm.org/viewvc/llvm-project/llvm/trunk/CODE_OWNERS.TXT?view=markup>`_
176 in the root of the LLVM source tree.
178 Note that code ownership is completely different than reviewers: anyone can
179 review a piece of code, and we welcome code review from anyone who is
180 interested. Code owners are the "last line of defense" to guarantee that all
181 patches that are committed are actually reviewed.
183 Being a code owner is a somewhat unglamorous position, but it is incredibly
184 important for the ongoing success of the project. Because people get busy,
185 interests change, and unexpected things happen, code ownership is purely opt-in,
186 and anyone can choose to resign their "title" at any time. For now, we do not
187 have an official policy on how one gets elected to be a code owner.
189 .. _include a testcase:
194 Developers are required to create test cases for any bugs fixed and any new
195 features added. Some tips for getting your testcase approved:
197 * All feature and regression test cases are added to the ``llvm/test``
198 directory. The appropriate sub-directory should be selected (see the
199 :doc:`Testing Guide <TestingGuide>` for details).
201 * Test cases should be written in `LLVM assembly language <LangRef.html>`_
202 unless the feature or regression being tested requires another language
203 (e.g. the bug being fixed or feature being implemented is in the llvm-gcc C++
204 front-end, in which case it must be written in C++).
206 * Test cases, especially for regressions, should be reduced as much as possible,
207 by `bugpoint <Bugpoint.html>`_ or manually. It is unacceptable to place an
208 entire failing program into ``llvm/test`` as this creates a *time-to-test*
209 burden on all developers. Please keep them short.
211 Note that llvm/test and clang/test are designed for regression and small feature
212 tests only. More extensive test cases (e.g., entire applications, benchmarks,
213 etc) should be added to the ``llvm-test`` test suite. The llvm-test suite is
214 for coverage (correctness, performance, etc) testing, not feature or regression
220 The minimum quality standards that any change must satisfy before being
221 committed to the main development branch are:
223 #. Code must adhere to the `LLVM Coding Standards <CodingStandards.html>`_.
225 #. Code must compile cleanly (no errors, no warnings) on at least one platform.
227 #. Bug fixes and new features should `include a testcase`_ so we know if the
228 fix/feature ever regresses in the future.
230 #. Code must pass the ``llvm/test`` test suite.
232 #. The code must not cause regressions on a reasonable subset of llvm-test,
233 where "reasonable" depends on the contributor's judgement and the scope of
234 the change (more invasive changes require more testing). A reasonable subset
235 might be something like "``llvm-test/MultiSource/Benchmarks``".
237 Additionally, the committer is responsible for addressing any problems found in
238 the future that the change is responsible for. For example:
240 * The code should compile cleanly on all supported platforms.
242 * The changes should not cause any correctness regressions in the ``llvm-test``
243 suite and must not cause any major performance regressions.
245 * The change set should not cause performance or correctness regressions for the
248 * The changes should not cause performance or correctness regressions in code
249 compiled by LLVM on all applicable targets.
251 * You are expected to address any `Bugzilla bugs <http://llvm.org/bugs/>`_ that
252 result from your change.
254 We prefer for this to be handled before submission but understand that it isn't
255 possible to test all of this for every submission. Our build bots and nightly
256 testing infrastructure normally finds these problems. A good rule of thumb is
257 to check the nightly testers for regressions the day after your change. Build
258 bots will directly email you if a group of commits that included yours caused a
259 failure. You are expected to check the build bot messages to see if they are
260 your fault and, if so, fix the breakage.
262 Commits that violate these quality standards (e.g. are very broken) may be
263 reverted. This is necessary when the change blocks other developers from making
264 progress. The developer is welcome to re-commit the change after the problem has
267 Obtaining Commit Access
268 -----------------------
270 We grant commit access to contributors with a track record of submitting high
271 quality patches. If you would like commit access, please send an email to
272 `Chris <mailto:sabre@nondot.org>`_ with the following information:
274 #. The user name you want to commit with, e.g. "hacker".
276 #. The full name and email address you want message to llvm-commits to come
277 from, e.g. "J. Random Hacker <hacker@yoyodyne.com>".
279 #. A "password hash" of the password you want to use, e.g. "``2ACR96qjUqsyM``".
280 Note that you don't ever tell us what your password is; you just give it to
281 us in an encrypted form. To get this, run "``htpasswd``" (a utility that
282 comes with apache) in crypt mode (often enabled with "``-d``"), or find a web
283 page that will do it for you.
285 Once you've been granted commit access, you should be able to check out an LLVM
286 tree with an SVN URL of "https://username@llvm.org/..." instead of the normal
287 anonymous URL of "http://llvm.org/...". The first time you commit you'll have
288 to type in your password. Note that you may get a warning from SVN about an
289 untrusted key; you can ignore this. To verify that your commit access works,
290 please do a test commit (e.g. change a comment or add a blank line). Your first
291 commit to a repository may require the autogenerated email to be approved by a
292 mailing list. This is normal and will be done when the mailing list owner has
295 If you have recently been granted commit access, these policies apply:
297 #. You are granted *commit-after-approval* to all parts of LLVM. To get
298 approval, submit a `patch`_ to `llvm-commits
299 <http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits>`_. When approved,
300 you may commit it yourself.
302 #. You are allowed to commit patches without approval which you think are
303 obvious. This is clearly a subjective decision --- we simply expect you to
304 use good judgement. Examples include: fixing build breakage, reverting
305 obviously broken patches, documentation/comment changes, any other minor
308 #. You are allowed to commit patches without approval to those portions of LLVM
309 that you have contributed or maintain (i.e., have been assigned
310 responsibility for), with the proviso that such commits must not break the
311 build. This is a "trust but verify" policy, and commits of this nature are
312 reviewed after they are committed.
314 #. Multiple violations of these policies or a single egregious violation may
315 cause commit access to be revoked.
317 In any case, your changes are still subject to `code review`_ (either before or
318 after they are committed, depending on the nature of the change). You are
319 encouraged to review other peoples' patches as well, but you aren't required
322 .. _discuss the change/gather consensus:
324 Making a Major Change
325 ---------------------
327 When a developer begins a major new project with the aim of contributing it back
328 to LLVM, s/he should inform the community with an email to the `llvmdev
329 <http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>`_ email list, to the extent
330 possible. The reason for this is to:
332 #. keep the community informed about future changes to LLVM,
334 #. avoid duplication of effort by preventing multiple parties working on the
335 same thing and not knowing about it, and
337 #. ensure that any technical issues around the proposed work are discussed and
338 resolved before any significant work is done.
340 The design of LLVM is carefully controlled to ensure that all the pieces fit
341 together well and are as consistent as possible. If you plan to make a major
342 change to the way LLVM works or want to add a major new extension, it is a good
343 idea to get consensus with the development community before you start working on
346 Once the design of the new feature is finalized, the work itself should be done
347 as a series of `incremental changes`_, not as a long-term development branch.
349 .. _incremental changes:
351 Incremental Development
352 -----------------------
354 In the LLVM project, we do all significant changes as a series of incremental
355 patches. We have a strong dislike for huge changes or long-term development
356 branches. Long-term development branches have a number of drawbacks:
358 #. Branches must have mainline merged into them periodically. If the branch
359 development and mainline development occur in the same pieces of code,
360 resolving merge conflicts can take a lot of time.
362 #. Other people in the community tend to ignore work on branches.
364 #. Huge changes (produced when a branch is merged back onto mainline) are
365 extremely difficult to `code review`_.
367 #. Branches are not routinely tested by our nightly tester infrastructure.
369 #. Changes developed as monolithic large changes often don't work until the
370 entire set of changes is done. Breaking it down into a set of smaller
371 changes increases the odds that any of the work will be committed to the main
374 To address these problems, LLVM uses an incremental development style and we
375 require contributors to follow this practice when making a large/invasive
378 * Large/invasive changes usually have a number of secondary changes that are
379 required before the big change can be made (e.g. API cleanup, etc). These
380 sorts of changes can often be done before the major change is done,
381 independently of that work.
383 * The remaining inter-related work should be decomposed into unrelated sets of
384 changes if possible. Once this is done, define the first increment and get
385 consensus on what the end goal of the change is.
387 * Each change in the set can be stand alone (e.g. to fix a bug), or part of a
388 planned series of changes that works towards the development goal.
390 * Each change should be kept as small as possible. This simplifies your work
391 (into a logical progression), simplifies code review and reduces the chance
392 that you will get negative feedback on the change. Small increments also
393 facilitate the maintenance of a high quality code base.
395 * Often, an independent precursor to a big change is to add a new API and slowly
396 migrate clients to use the new API. Each change to use the new API is often
397 "obvious" and can be committed without review. Once the new API is in place
398 and used, it is much easier to replace the underlying implementation of the
399 API. This implementation change is logically separate from the API
402 If you are interested in making a large change, and this scares you, please make
403 sure to first `discuss the change/gather consensus`_ then ask about the best way
404 to go about making the change.
406 Attribution of Changes
407 ----------------------
409 We believe in correct attribution of contributions to their contributors.
410 However, we do not want the source code to be littered with random attributions
411 "this code written by J. Random Hacker" (this is noisy and distracting). In
412 practice, the revision control system keeps a perfect history of who changed
413 what, and the CREDITS.txt file describes higher-level contributions. If you
414 commit a patch for someone else, please say "patch contributed by J. Random
415 Hacker!" in the commit message.
417 Overall, please do not add contributor names to the source code.
419 .. _copyright-license-patents:
421 Copyright, License, and Patents
422 ===============================
426 This section deals with legal matters but does not provide legal advice. We
427 are not lawyers --- please seek legal counsel from an attorney.
429 This section addresses the issues of copyright, license and patents for the LLVM
430 project. The copyright for the code is held by the individual contributors of
431 the code and the terms of its license to LLVM users and developers is the
432 `University of Illinois/NCSA Open Source License
433 <http://www.opensource.org/licenses/UoI-NCSA.php>`_ (with portions dual licensed
434 under the `MIT License <http://www.opensource.org/licenses/mit-license.php>`_,
435 see below). As contributor to the LLVM project, you agree to allow any
436 contributions to the project to licensed under these terms.
441 The LLVM project does not require copyright assignments, which means that the
442 copyright for the code in the project is held by its respective contributors who
443 have each agreed to release their contributed code under the terms of the `LLVM
446 An implication of this is that the LLVM license is unlikely to ever change:
447 changing it would require tracking down all the contributors to LLVM and getting
448 them to agree that a license change is acceptable for their contribution. Since
449 there are no plans to change the license, this is not a cause for concern.
451 As a contributor to the project, this means that you (or your company) retain
452 ownership of the code you contribute, that it cannot be used in a way that
453 contradicts the license (which is a liberal BSD-style license), and that the
454 license for your contributions won't change without your approval in the
462 We intend to keep LLVM perpetually open source and to use a liberal open source
463 license. **As a contributor to the project, you agree that any contributions be
464 licensed under the terms of the corresponding subproject.** All of the code in
465 LLVM is available under the `University of Illinois/NCSA Open Source License
466 <http://www.opensource.org/licenses/UoI-NCSA.php>`_, which boils down to
469 * You can freely distribute LLVM.
470 * You must retain the copyright notice if you redistribute LLVM.
471 * Binaries derived from LLVM must reproduce the copyright notice (e.g. in an
472 included readme file).
473 * You can't use our names to promote your LLVM derived products.
474 * There's no warranty on LLVM at all.
476 We believe this fosters the widest adoption of LLVM because it **allows
477 commercial products to be derived from LLVM** with few restrictions and without
478 a requirement for making any derived works also open source (i.e. LLVM's
479 license is not a "copyleft" license like the GPL). We suggest that you read the
480 `License <http://www.opensource.org/licenses/UoI-NCSA.php>`_ if further
481 clarification is needed.
483 In addition to the UIUC license, the runtime library components of LLVM
484 (**compiler_rt, libc++, and libclc**) are also licensed under the `MIT License
485 <http://www.opensource.org/licenses/mit-license.php>`_, which does not contain
486 the binary redistribution clause. As a user of these runtime libraries, it
487 means that you can choose to use the code under either license (and thus don't
488 need the binary redistribution clause), and as a contributor to the code that
489 you agree that any contributions to these libraries be licensed under both
490 licenses. We feel that this is important for runtime libraries, because they
491 are implicitly linked into applications and therefore should not subject those
492 applications to the binary redistribution clause. This also means that it is ok
493 to move code from (e.g.) libc++ to the LLVM core without concern, but that code
494 cannot be moved from the LLVM core to libc++ without the copyright owner's
497 Note that the LLVM Project does distribute llvm-gcc and dragonegg, **which are
498 GPL.** This means that anything "linked" into llvm-gcc must itself be compatible
499 with the GPL, and must be releasable under the terms of the GPL. This implies
500 that **any code linked into llvm-gcc and distributed to others may be subject to
501 the viral aspects of the GPL** (for example, a proprietary code generator linked
502 into llvm-gcc must be made available under the GPL). This is not a problem for
503 code already distributed under a more liberal license (like the UIUC license),
504 and GPL-containing subprojects are kept in separate SVN repositories whose
505 LICENSE.txt files specifically indicate that they contain GPL code.
507 We have no plans to change the license of LLVM. If you have questions or
508 comments about the license, please contact the `LLVM Developer's Mailing
509 List <mailto:llvmdev@cs.uiuc.edu>`_.
514 To the best of our knowledge, LLVM does not infringe on any patents (we have
515 actually removed code from LLVM in the past that was found to infringe). Having
516 code in LLVM that infringes on patents would violate an important goal of the
517 project by making it hard or impossible to reuse the code for arbitrary purposes
518 (including commercial use).
520 When contributing code, we expect contributors to notify us of any potential for
521 patent-related trouble with their changes (including from third parties). If
522 you or your employer own the rights to a patent and would like to contribute
523 code to LLVM that relies on it, we require that the copyright owner sign an
524 agreement that allows any other user of LLVM to freely use your patent. Please
525 contact the `oversight group <mailto:llvm-oversight@cs.uiuc.edu>`_ for more