| blob | fee28384ebb163353252d6d4ba830056a611c577 |
3 <html>
4 <head>
8 </head>
9 <body>
12 <ol>
15 <ol>
26 </ol></li>
28 <ol>
32 </ol></li>
33 </ol>
36 <!--=========================================================================-->
38 <!--=========================================================================-->
39 <div>
40 <p>This document contains the LLVM Developer Policy which defines the project's
41 policy towards developers and their contributions. The intent of this policy
42 is to eliminate miscommunication, rework, and confusion that might arise from
43 the distributed nature of LLVM's development. By stating the policy in clear
44 terms, we hope each developer can know ahead of time what to expect when
45 making LLVM contributions. This policy covers all llvm.org subprojects,
46 including Clang, LLDB, etc.</p>
49 <ol>
55 </ol>
57 <p>This policy is aimed at frequent contributors to LLVM. People interested in
58 contributing one-off patches can do so in an informal way by sending them to
59 the
61 mailing list</a> and engaging another developer to see it through the
62 process.</p>
63 </div>
65 <!--=========================================================================-->
67 <!--=========================================================================-->
68 <div>
69 <p>This section contains policies that pertain to frequent LLVM developers. We
71 routinely contribute to LLVM, but we expect more from frequent contributors
72 to keep the system as efficient as possible for everyone. Frequent LLVM
73 contributors are expected to meet the following requirements in order for
74 LLVM to maintain a high standard of quality.<p>
76 <!-- _______________________________________________________________________ -->
78 <div>
80 for the projects you are interested in, such as
83 for Clang, or <a
85 for LLDB. If you are doing anything more than just casual work on LLVM, it
86 is suggested that you also subscribe to the "commits" mailing list for the
87 subproject you're interested in, such as
91 Reading the "commits" list and paying attention to changes being made by
92 others is a good way to see what other people are interested in and watching
93 the flow of the project as a whole.</p>
95 <p>We recommend that active developers register an email account with
98 email list to keep track of bugs and enhancements occurring in LLVM. We
99 really appreciate people who are proactive at catching incoming bugs in their
100 components and dealing with them promptly.</p>
101 </div>
103 <!-- _______________________________________________________________________ -->
106 <div>
107 <p>When making a patch for review, the goal is to make it as easy for the
108 reviewer to read it as possible. As such, we recommend that you:</p>
110 <ol>
111 <li>Make your patch against the Subversion trunk, not a branch, and not an old
112 version of LLVM. This makes it easy to apply the patch. For information
113 on how to check out SVN trunk, please see the <a
116 <li>Similarly, patches should be submitted soon after they are generated. Old
117 patches may not apply correctly if the underlying code changes between the
118 time the patch was created and the time it is applied.</li>
122 that it doesn't contain clutter which makes it hard to read.</li>
124 <li>If you are modifying generated files, such as the top-level
126 a separate patch from the rest of your changes.</li>
127 </ol>
129 <p>When sending a patch to a mailing list, it is a good idea to send it as an
131 message. This ensures that your mailer will not mangle the patch when it
132 sends it (e.g. by making whitespace changes or by wrapping lines).</p>
139 attachment</tt>. Apple Mail gamely displays such a file inline, making it
140 difficult to work with for reviewers using that program.</p>
141 </div>
143 <!-- _______________________________________________________________________ -->
145 <div>
146 <p>LLVM has a code review policy. Code review is one way to increase the quality
147 of software. We generally follow these policies:</p>
149 <ol>
150 <li>All developers are required to have significant changes reviewed before
151 they are committed to the repository.</li>
153 <li>Code reviews are conducted by email, usually on the llvm-commits
154 list.</li>
156 <li>Code can be reviewed either before it is committed or after. We expect
157 major changes to be reviewed before being committed, but smaller changes
158 (or changes where the developer owns the component) can be reviewed after
159 commit.</li>
161 <li>The developer responsible for a code change is also responsible for making
162 all necessary review-related changes.</li>
164 <li>Code review can be an iterative process, which continues until the patch
165 is ready to be committed.</li>
166 </ol>
168 <p>Developers should participate in code reviews as both reviewers and
169 reviewees. If someone is kind enough to review your code, you should return
170 the favor for someone else. Note that anyone is welcome to review and give
171 feedback on a patch, but only people with Subversion write access can approve
172 it.</p>
173 </div>
175 <!-- _______________________________________________________________________ -->
177 <div>
179 <p>The LLVM Project relies on two features of its process to maintain rapid
180 development in addition to the high quality of its source base: the
181 combination of code review plus post-commit review for trusted maintainers.
182 Having both is a great way for the project to take advantage of the fact that
183 most people do the right thing most of the time, and only commit patches
184 without pre-commit review when they are confident they are right.</p>
186 <p>The trick to this is that the project has to guarantee that all patches that
187 are committed are reviewed after they go in: you don't want everyone to
188 assume someone else will review it, allowing the patch to go unreviewed. To
189 solve this problem, we have a notion of an 'owner' for a piece of the code.
190 The sole responsibility of a code owner is to ensure that a commit to their
191 area of the code is appropriately reviewed, either by themself or by someone
192 else. The current code owners are:</p>
194 <ol>
204 Windows codegen.</li>
211 </ol>
213 <p>Note that code ownership is completely different than reviewers: anyone can
214 review a piece of code, and we welcome code review from anyone who is
215 interested. Code owners are the "last line of defense" to guarantee that all
216 patches that are committed are actually reviewed.</p>
218 <p>Being a code owner is a somewhat unglamorous position, but it is incredibly
219 important for the ongoing success of the project. Because people get busy,
220 interests change, and unexpected things happen, code ownership is purely
221 opt-in, and anyone can choose to resign their "title" at any time. For now,
222 we do not have an official policy on how one gets elected to be a code
223 owner.</p>
224 </div>
226 <!-- _______________________________________________________________________ -->
228 <div>
229 <p>Developers are required to create test cases for any bugs fixed and any new
230 features added. Some tips for getting your testcase approved:</p>
232 <ol>
233 <li>All feature and regression test cases are added to the
236 details).</li>
239 language</a> unless the feature or regression being tested requires
240 another language (e.g. the bug being fixed or feature being implemented is
241 in the llvm-gcc C++ front-end, in which case it must be written in
242 C++).</li>
244 <li>Test cases, especially for regressions, should be reduced as much as
248 them short.</li>
249 </ol>
251 <p>Note that llvm/test and clang/test are designed for regression and small
252 feature tests only. More extensive test cases (e.g., entire applications,
253 benchmarks, etc)
255 for coverage (correctness, performance, etc) testing, not feature or
256 regression testing.</p>
257 </div>
259 <!-- _______________________________________________________________________ -->
261 <div>
262 <p>The minimum quality standards that any change must satisfy before being
263 committed to the main development branch are:</p>
265 <ol>
269 <li>Code must compile cleanly (no errors, no warnings) on at least one
270 platform.</li>
273 testcase</a> so we know if the fix/feature ever regresses in the
274 future.</li>
278 <li>The code must not cause regressions on a reasonable subset of llvm-test,
279 where "reasonable" depends on the contributor's judgement and the scope of
280 the change (more invasive changes require more testing). A reasonable
281 subset might be something like
283 </ol>
285 <p>Additionally, the committer is responsible for addressing any problems found
286 in the future that the change is responsible for. For example:</p>
288 <ul>
291 <li>The changes should not cause any correctness regressions in the
293 regressions.</li>
295 <li>The change set should not cause performance or correctness regressions for
296 the LLVM tools.</li>
298 <li>The changes should not cause performance or correctness regressions in
299 code compiled by LLVM on all applicable targets.</li>
303 </ul>
305 <p>We prefer for this to be handled before submission but understand that it
306 isn't possible to test all of this for every submission. Our build bots and
307 nightly testing infrastructure normally finds these problems. A good rule of
308 thumb is to check the nightly testers for regressions the day after your
309 change. Build bots will directly email you if a group of commits that
310 included yours caused a failure. You are expected to check the build bot
311 messages to see if they are your fault and, if so, fix the breakage.</p>
313 <p>Commits that violate these quality standards (e.g. are very broken) may be
314 reverted. This is necessary when the change blocks other developers from
315 making progress. The developer is welcome to re-commit the change after the
316 problem has been fixed.</p>
317 </div>
319 <!-- _______________________________________________________________________ -->
321 <div>
323 <p>We grant commit access to contributors with a track record of submitting high
324 quality patches. If you would like commit access, please send an email to
326 information:</p>
328 <ol>
331 <li>The full name and email address you want message to llvm-commits to come
335 Note that you don't ever tell us what your password is, you just give it
336 to us in an encrypted form. To get this, run "htpasswd" (a utility that
337 comes with apache) in crypt mode (often enabled with "-d"), or find a web
338 page that will do it for you.</li>
339 </ol>
341 <p>Once you've been granted commit access, you should be able to check out an
342 LLVM tree with an SVN URL of "https://username@llvm.org/..." instead of the
343 normal anonymous URL of "http://llvm.org/...". The first time you commit
344 you'll have to type in your password. Note that you may get a warning from
345 SVN about an untrusted key, you can ignore this. To verify that your commit
346 access works, please do a test commit (e.g. change a comment or add a blank
347 line). Your first commit to a repository may require the autogenerated email
348 to be approved by a mailing list. This is normal, and will be done when
349 the mailing list owner has time.</p>
353 <ol>
357 When approved you may commit it yourself.</li>
359 <li>You are allowed to commit patches without approval which you think are
360 obvious. This is clearly a subjective decision — we simply expect
361 you to use good judgement. Examples include: fixing build breakage,
362 reverting obviously broken patches, documentation/comment changes, any
363 other minor changes.</li>
365 <li>You are allowed to commit patches without approval to those portions of
366 LLVM that you have contributed or maintain (i.e., have been assigned
367 responsibility for), with the proviso that such commits must not break the
368 build. This is a "trust but verify" policy and commits of this nature are
369 reviewed after they are committed.</li>
371 <li>Multiple violations of these policies or a single egregious violation may
372 cause commit access to be revoked.</li>
373 </ol>
376 review</a> (either before or after they are committed, depending on the
377 nature of the change). You are encouraged to review other peoples' patches
378 as well, but you aren't required to.</p>
379 </div>
381 <!-- _______________________________________________________________________ -->
383 <div>
384 <p>When a developer begins a major new project with the aim of contributing it
385 back to LLVM, s/he should inform the community with an email to
387 email list, to the extent possible. The reason for this is to:
389 <ol>
392 <li>avoid duplication of effort by preventing multiple parties working on the
393 same thing and not knowing about it, and</li>
395 <li>ensure that any technical issues around the proposed work are discussed
396 and resolved before any significant work is done.</li>
397 </ol>
399 <p>The design of LLVM is carefully controlled to ensure that all the pieces fit
400 together well and are as consistent as possible. If you plan to make a major
401 change to the way LLVM works or want to add a major new extension, it is a
402 good idea to get consensus with the development community before you start
403 working on it.</p>
405 <p>Once the design of the new feature is finalized, the work itself should be
407 long-term development branch.</p>
408 </div>
410 <!-- _______________________________________________________________________ -->
412 <div>
413 <p>In the LLVM project, we do all significant changes as a series of incremental
414 patches. We have a strong dislike for huge changes or long-term development
415 branches. Long-term development branches have a number of drawbacks:</p>
417 <ol>
418 <li>Branches must have mainline merged into them periodically. If the branch
419 development and mainline development occur in the same pieces of code,
420 resolving merge conflicts can take a lot of time.</li>
424 <li>Huge changes (produced when a branch is merged back onto mainline) are
427 <li>Branches are not routinely tested by our nightly tester
428 infrastructure.</li>
430 <li>Changes developed as monolithic large changes often don't work until the
431 entire set of changes is done. Breaking it down into a set of smaller
432 changes increases the odds that any of the work will be committed to the
433 main repository.</li>
434 </ol>
436 <p>To address these problems, LLVM uses an incremental development style and we
437 require contributors to follow this practice when making a large/invasive
438 change. Some tips:</p>
440 <ul>
441 <li>Large/invasive changes usually have a number of secondary changes that are
442 required before the big change can be made (e.g. API cleanup, etc). These
443 sorts of changes can often be done before the major change is done,
444 independently of that work.</li>
446 <li>The remaining inter-related work should be decomposed into unrelated sets
447 of changes if possible. Once this is done, define the first increment and
448 get consensus on what the end goal of the change is.</li>
450 <li>Each change in the set can be stand alone (e.g. to fix a bug), or part of
451 a planned series of changes that works towards the development goal.</li>
453 <li>Each change should be kept as small as possible. This simplifies your work
454 (into a logical progression), simplifies code review and reduces the
455 chance that you will get negative feedback on the change. Small increments
456 also facilitate the maintenance of a high quality code base.</li>
458 <li>Often, an independent precursor to a big change is to add a new API and
459 slowly migrate clients to use the new API. Each change to use the new API
460 is often "obvious" and can be committed without review. Once the new API
461 is in place and used, it is much easier to replace the underlying
462 implementation of the API. This implementation change is logically
463 separate from the API change.</li>
464 </ul>
466 <p>If you are interested in making a large change, and this scares you, please
468 then ask about the best way to go about making the change.</p>
469 </div>
471 <!-- _______________________________________________________________________ -->
473 <div>
474 <p>We believe in correct attribution of contributions to their contributors.
475 However, we do not want the source code to be littered with random
476 attributions "this code written by J. Random Hacker" (this is noisy and
477 distracting). In practice, the revision control system keeps a perfect
478 history of who changed what, and the CREDITS.txt file describes higher-level
479 contributions. If you commit a patch for someone else, please say "patch
483 </div>
485 </div>
487 <!--=========================================================================-->
488 <h2>
490 </h2>
491 <!--=========================================================================-->
493 <div>
494 <p>This section addresses the issues of copyright, license and patents for the
495 LLVM project. Currently, the University of Illinois is the LLVM copyright
496 holder and the terms of its license to LLVM users and developers is the
502 legal matters but does not provide legal advice. We are not lawyers, please
503 seek legal counsel from an attorney.</p>
504 </div>
506 <!-- _______________________________________________________________________ -->
508 <div>
510 <p>The LLVM project does not require copyright assignments, which means that the
511 copyright for the code in the project is held by its respective contributors
512 who have each agreed to release their contributed code under the terms of the
515 <p>An implication of this is that the LLVM license is unlikely to ever change:
516 changing it would require tracking down all the contributors to LLVM and
517 getting them to agree that a license change is acceptable for their
518 contribution. Since there are no plans to change the license, this is not a
519 cause for concern.</p>
521 <p>As a contributor to the project, this means that you (or your company) retain
522 ownership of the code you contribute, that it cannot be used in a way that
523 contradicts the license (which is a liberal BSD-style license), and that the
524 license for your contributions won't change without your approval in the
525 future.</p>
527 </div>
529 <!-- _______________________________________________________________________ -->
531 <div>
532 <p>We intend to keep LLVM perpetually open source and to use a liberal open
533 source license. All of the code in LLVM is available under the
537 <ul>
540 <li>Binaries derived from LLVM must reproduce the copyright notice (e.g. in an
541 included readme file).</li>
544 </ul>
547 commercial products to be derived from LLVM</b> with few restrictions and
548 without a requirement for making any derived works also open source (i.e.
549 LLVM's license is not a "copyleft" license like the GPL). We suggest that you
551 if further clarification is needed.</p>
553 <p>In addition to the UIUC license, the runtime library components of LLVM
556 which does not contain the binary redistribution clause. As a user of these
557 runtime libraries, it means that you can choose to use the code under either
558 license (and thus don't need the binary redistribution clause), and as a
559 contributor to the code that you agree that any contributions to these
560 libraries be licensed under both licenses. We feel that this is important
561 for runtime libraries, because they are implicitly linked into applications
562 and therefore should not subject those applications to the binary
563 redistribution clause. This also means that it is ok to move code from (e.g.)
564 libc++ to the LLVM core without concern, but that code cannot be moved from
565 the LLVM core to libc++ without the copyright owner's permission.
566 </p>
569 This means that anything "linked" into llvm-gcc must itself be compatible
570 with the GPL, and must be releasable under the terms of the GPL. This
571 implies that <b>any code linked into llvm-gcc and distributed to others may
572 be subject to the viral aspects of the GPL</b> (for example, a proprietary
573 code generator linked into llvm-gcc must be made available under the GPL).
574 This is not a problem for code already distributed under a more liberal
575 license (like the UIUC license), and does not affect code generated by
576 llvm-gcc. It may be a problem if you intend to base commercial development
577 on llvm-gcc without redistributing your source code.</p>
579 <p>We have no plans to change the license of LLVM. If you have questions or
580 comments about the license, please contact the
582 </div>
584 <!-- _______________________________________________________________________ -->
586 <div>
587 <p>To the best of our knowledge, LLVM does not infringe on any patents (we have
588 actually removed code from LLVM in the past that was found to infringe).
589 Having code in LLVM that infringes on patents would violate an important goal
590 of the project by making it hard or impossible to reuse the code for
591 arbitrary purposes (including commercial use).</p>
593 <p>When contributing code, we expect contributors to notify us of any potential
594 for patent-related trouble with their changes. If you or your employer own
595 the rights to a patent and would like to contribute code to LLVM that relies
596 on it, we require that the copyright owner sign an agreement that allows any
597 other user of LLVM to freely use your patent. Please contact
599 details.</p>
600 </div>
602 </div>
604 <!-- *********************************************************************** -->
605 <hr>
606 <address>
611 Written by the
614 Last modified: $Date$
615 </address>
616 </body>
617 </html>