| blob | 18c1ff34ac273e6aa53550bc1aff69be0a932b44 |
3 <html>
4 <head>
7 </head>
8 <body>
11 <ol>
14 <ol>
25 </ol></li>
27 <ol>
32 </ol></li>
33 </ol>
36 <!--=========================================================================-->
38 <!--=========================================================================-->
40 <p>This document contains the LLVM Developer Policy which defines the
41 project's policy towards developers and their contributions. The intent of
42 this policy is to eliminate mis-communication, rework, and confusion that
43 might arise from the distributed nature of LLVM's development. By stating
44 the policy in clear terms, we hope each developer can know ahead of time
45 what to expect when making LLVM contributions.</p>
47 <ol>
51 </ol>
53 <p>This policy is aimed at frequent contributors to LLVM. People interested in
54 contributing one-off patches can do so in an informal way by sending them to
56 llvm-commits mailing list</a> and engaging another developer to see it through
57 the process.</p>
59 </div>
61 <!--=========================================================================-->
63 <!--=========================================================================-->
65 <p>This section contains policies that pertain to frequent LLVM
67 people who do not routinely contribute to LLVM, but we expect more from
68 frequent contributors to keep the system as efficient as possible for
69 everyone.
70 Frequent LLVM contributors are expected to meet the following requirements in
71 order for LLVM to maintain a high standard of quality.<p>
72 </div>
74 <!-- _______________________________________________________________________ -->
77 <p>Developers should stay informed by reading at least the
79 email list. If you are doing anything more than just casual work on LLVM,
80 it is suggested that you also subscribe to the
82 list and pay attention to changes being made by others.</p>
83 <p>We recommend that active developers register an email account with
86 email list to keep track of bugs and enhancements occurring in LLVM.</p>
87 </div>
89 <!-- _______________________________________________________________________ -->
94 <p>When making a patch for review, the goal is to make it as easy for the
95 reviewer to read it as possible. As such, we recommend that you:</p>
96 <ol>
97 <li>Make your patch against the Subversion trunk, not a branch, and not an
98 old version of LLVM. This makes it easy to apply the patch.</li>
100 <li>Similarly, patches should be submitted soon after they are generated.
101 Old patches may not apply correctly if the underlying code changes between
102 the time the patch was created and the time it is applied.</li>
104 <li>Patches should be made with this command:
107 diff.</li>
109 <li>Patches should not include differences in generated code such as the
113 </ol>
115 <p>When sending a patch to a mailing list, it is a good idea to send it as an
117 message. This ensures that your mailer will not mangle the patch when it
118 sends it (e.g. by making whitespace changes or by wrapping lines).</p>
119 </div>
121 <!-- _______________________________________________________________________ -->
124 <p>LLVM has a code review policy. Code review is one way to increase the
125 quality of software. We generally follow these policies:</p>
126 <ol>
127 <li>All developers are required to have significant changes reviewed
128 before they are committed to the repository.</li>
129 <li>Code reviews are conducted by email, usually on the llvm-commits
130 list.</li>
131 <li>Code can be reviewed either before it is committed or after. We expect
132 major changes to be reviewed before being committed, but smaller
133 changes (or changes where the developer owns the component) can be
134 reviewed after commit.</li>
135 <li>The developer responsible for a code change is also responsible for
136 making all necessary review-related changes.</li>
137 <li>Code review can be an iterative process, which continues until the patch
138 is ready to be committed.</li>
139 </ol>
141 <p>Developers should participate in code reviews as both reviewers and
142 reviewees. If someone is kind enough to review your code, you should
143 return the favor for someone else. Note that anyone is welcome to review
144 and give feedback on a patch, but only people with Subversion write access
145 can approve it.</p>
147 </div>
149 <!-- _______________________________________________________________________ -->
153 <p>The LLVM Project relies on two features of its process to maintain rapid
154 development in addition to the high quality of its source base: the
155 combination of code review plus post-commit review for trusted maintainers.
156 Having both is a great way for the project to take advantage of the fact
157 that most people do the right thing most of the time, and only commit
158 patches without pre-commit review when they are confident they are
159 right.</p>
161 <p>The trick to this is that the project has to guarantee that all patches
162 that are committed are reviewed after they go in: you don't want everyone
163 to assume someone else will review it, allowing the patch to go unreviewed.
164 To solve this problem, we have a notion of an 'owner' for a piece of the
165 code. The sole responsibility of a code owner is to ensure that a commit
166 to their area of the code is appropriately reviewed, either by themself or
167 by someone else. The current code owners are:</p>
169 <ol>
171 Windows codegen.</li>
175 </ol>
177 <p>Note that code ownership is completely different than reviewers: anyone can
178 review a piece of code, and we welcome code review from anyone who is
179 interested. Code owners are the "last line of defense" to guarantee that
180 all patches that are committed are actually reviewed.</p>
182 <p>Being a code owner is a somewhat unglamorous position, but it is incredibly
183 important for the ongoing success of the project. Because people get busy,
184 interests change, and unexpected things happen, code ownership is purely
185 opt-in, and anyone can choose to resign their "title" at any time. For now,
186 we do not have an official policy on how one gets elected to be a code
187 owner.
188 </p>
190 </div>
193 <!-- _______________________________________________________________________ -->
196 <p>Developers are required to create test cases for any bugs fixed and any new
197 features added. Some tips for getting your testcase approved:</p>
198 <ol>
199 <li>All feature and regression test cases are added to the
202 details).</li>
203 <li>Test cases should be written in
205 feature or regression being tested requires another language (e.g. the
206 bug being fixed or feature being implemented is in the llvm-gcc C++
207 front-end, in which case it must be written in C++).</li>
208 <li>Test cases, especially for regressions, should be reduced as much as
210 manually. It is unacceptable
213 </ol>
215 <p>Note that llvm/test is designed for regression and small feature tests
216 only. More extensive test cases (e.g., entire applications, benchmarks,
218 suite is for coverage (correctness, performance, etc) testing, not feature
219 or regression testing.</p>
220 </div>
222 <!-- _______________________________________________________________________ -->
225 <p>The minimum quality standards that any change must satisfy before being
226 committed to the main development branch are:</p>
227 <ol>
228 <li>Code must adhere to the
230 <li>Code must compile cleanly (no errors, no warnings) on at least one
231 platform.</li>
233 testcase</a> so we know if the fix/feature ever regresses in the
234 future.</li>
236 <li>The code must not cause regressions on a reasonable subset of llvm-test,
237 where "reasonable" depends on the contributor's judgement and the scope
238 of the change (more invasive changes require more testing). A reasonable
240 </ol>
241 <p>Additionally, the committer is responsible for addressing any problems
242 found in the future that the change is responsible for. For example:</p>
243 <ul>
245 <li>The changes should not cause any correctness regressions in the
247 regressions.</li>
248 <li>The change set should not cause performance or correctness regressions
249 for the LLVM tools.</li>
250 <li>The changes should not cause performance or correctness regressions in
251 code compiled by LLVM on all applicable targets.</li>
254 </ul>
256 <p>We prefer for this to be handled before submission but understand that it
257 isn't possible to test all of this for every submission. Our nightly
258 testing
259 infrastructure normally finds these problems. A good rule of thumb is to
260 check the nightly testers for regressions the day after your change.</p>
262 <p>Commits that violate these quality standards (e.g. are very broken) may
263 be reverted. This is necessary when the change blocks other developers from
264 making progress. The developer is welcome to re-commit the change after
265 the problem has been fixed.</p>
266 </div>
268 <!-- _______________________________________________________________________ -->
273 <p>
274 We grant commit access to contributors with a track record of submitting high
275 quality patches. If you would like commit access, please send an email to
278 <ol>
280 <li>The full name and email address you want message to llvm-commits to come
283 Note that you don't ever tell us what your password is, you just give it
284 to us in an encrypted form. To get this, run "htpasswd" (a utility that
285 comes with apache) in crypt mode (often enabled with "-d"), or find a web
286 page that will do it for you.</li>
287 </ol>
289 <p>Once you've been granted commit access, you should be able to check out an
290 LLVM tree with an SVN URL of "https://username@llvm.org/..." instead of the
291 normal anonymous URL of "http://llvm.org/...". The first time you commit
292 you'll have to type in your password. Note that you may get a warning from
293 SVN about an untrusted key, you can ignore this. To verify that your commit
294 access works, please do a test commit (e.g. change a comment or add a blank
295 line). Your first commit to a repository may require the autogenerated email
296 to be approved by a mailing list. This is normal, and will be done when
297 the mailing list owner has time.</p>
301 <ol>
306 <li>You are allowed to commit patches without approval which you think are
307 obvious. This is clearly a subjective decision — we simply expect you
308 to use good judgement. Examples include: fixing build breakage, reverting
309 obviously broken patches, documentation/comment changes, any other minor
310 changes.</li>
311 <li>You are allowed to commit patches without approval to those portions
312 of LLVM that you have contributed or maintain (i.e., have been assigned
313 responsibility for), with the proviso that such commits must not break the
314 build. This is a "trust but verify" policy and commits of this nature are
315 reviewed after they are committed.</li>
316 <li>Multiple violations of these policies or a single egregious violation
317 may cause commit access to be revoked.</li>
318 </ol>
321 review</a> (either before or after they are committed, depending on the nature
322 of the change). You are encouraged to review other peoples' patches as well,
323 but you aren't required to.</p>
325 </div>
327 <!-- _______________________________________________________________________ -->
330 <p>When a developer begins a major new project with the aim of contributing
331 it back to LLVM, s/he should inform the community with an email to
333 email list, to the extent possible. The reason for this is to:
334 <ol>
336 <li>avoid duplication of effort by preventing multiple parties working on
337 the same thing and not knowing about it, and</li>
338 <li>ensure that any technical issues around the proposed work are
339 discussed and resolved before any significant work is done.</li>
340 </ol>
342 <p>The design of LLVM is carefully controlled to ensure that all the pieces
343 fit together well and are as consistent as possible. If you plan to make a
344 major change to the way LLVM works or want to add a major new extension, it
345 is a good idea to get consensus with the development
346 community before you start working on it.</p>
348 <p>Once the design of the new feature is finalized, the work itself should be
350 a long-term development branch.</p>
352 </div>
354 <!-- _______________________________________________________________________ -->
356 </div>
358 <p>In the LLVM project, we do all significant changes as a series of
359 incremental patches. We have a strong dislike for huge changes or
360 long-term development branches. Long-term development branches have a
361 number of drawbacks:</p>
363 <ol>
364 <li>Branches must have mainline merged into them periodically. If the branch
365 development and mainline development occur in the same pieces of code,
366 resolving merge conflicts can take a lot of time.</li>
368 <li>Huge changes (produced when a branch is merged back onto mainline) are
370 <li>Branches are not routinely tested by our nightly tester
371 infrastructure.</li>
372 <li>Changes developed as monolithic large changes often don't work until the
373 entire set of changes is done. Breaking it down into a set of smaller
374 changes increases the odds that any of the work will be committed to the
375 main repository.</li>
376 </ol>
378 <p>
379 To address these problems, LLVM uses an incremental development style and we
380 require contributors to follow this practice when making a large/invasive
381 change. Some tips:</p>
383 <ul>
384 <li>Large/invasive changes usually have a number of secondary changes that
385 are required before the big change can be made (e.g. API cleanup, etc).
386 These sorts of changes can often be done before the major change is done,
387 independently of that work.</li>
388 <li>The remaining inter-related work should be decomposed into unrelated
389 sets of changes if possible. Once this is done, define the first increment
390 and get consensus on what the end goal of the change is.</li>
392 <li>Each change in the set can be stand alone (e.g. to fix a bug), or part
393 of a planned series of changes that works towards the development goal.</li>
395 <li>Each change should be kept as small as possible. This simplifies your
396 work (into a logical progression), simplifies code review and reduces the
397 chance that you will get negative feedback on the change. Small increments
398 also facilitate the maintenance of a high quality code base.</li>
400 <li>Often, an independent precursor to a big change is to add a new API and
401 slowly migrate clients to use the new API. Each change to use the new
402 API is often "obvious" and can be committed without review. Once the
403 new API is in place and used, it is much easier to replace the
404 underlying implementation of the API. This implementation change is
405 logically separate from the API change.</li>
406 </ul>
408 <p>If you are interested in making a large change, and this scares you, please
410 consensus</a> then ask about the best way to go about making
411 the change.</p>
412 </div>
414 <!-- _______________________________________________________________________ -->
416 Changes</a></div>
418 <p>We believe in correct attribution of contributions to
419 their contributors. However, we do not want the source code to be littered
420 with random attributions "this code written by J Random Guy" (this is noisy
421 and distracting. In practice, the revision control system keeps a perfect
422 history of who change what, and the CREDITS.txt file describes higher-level
423 contributions.</p>
426 </div>
430 <!--=========================================================================-->
433 </div>
434 <!--=========================================================================-->
437 <p>This section addresses the issues of copyright, license and patents for
438 the LLVM project.
439 Currently, the University of Illinois is the LLVM copyright holder and the
440 terms of its license to LLVM users and developers is the
445 <p><b>NOTE: This section deals with legal matters but does not provide
446 legal advice. We are not lawyers, please seek legal counsel from an
447 attorney.</b></p>
448 </div>
449 </div>
451 <!-- _______________________________________________________________________ -->
454 <p>
455 <p>For consistency and ease of management, the project requires the
456 copyright for all LLVM software to be held by a single copyright holder:
457 the University of Illinois (UIUC).</p>
459 <p>
460 Although UIUC may eventually reassign the copyright of the software to another
461 entity (e.g. a dedicated non-profit "LLVM Organization", or something)
462 the intent for the project is to always have a single entity hold the
463 copyrights to LLVM at any given time.</p>
465 <p>We believe that having a single copyright
466 holder is in the best interests of all developers and users as it greatly
467 reduces the managerial burden for any kind of administrative or technical
468 decisions about LLVM. The goal of the LLVM project is to always keep the code
470 </div>
472 <!-- _______________________________________________________________________ -->
475 <p>We intend to keep LLVM perpetually open source
476 and to use a liberal open source license. The current license is the
478 University of Illinois/NCSA Open Source License</a>, which boils
479 down to this:</p>
480 <ul>
486 </ul>
489 commercial products to be derived from LLVM</b> with few restrictions and
490 without a requirement for making any derived works also open source (i.e.
491 LLVM's license is not a "copyleft" license like the GPL). We suggest that you
493 if further clarification is needed.</p>
496 This means that anything "linked" into llvm-gcc must itself be compatible
497 with the GPL, and must be releasable under the terms of the GPL. This implies
498 that <b>any code linked into llvm-gcc and distributed to others may be subject
499 to the viral aspects of the GPL</b> (for example, a proprietary code generator
500 linked into llvm-gcc must be made available under the GPL). This is not a
501 problem for code already distributed under a more liberal license (like the
502 UIUC license), and does not affect code generated by llvm-gcc. It may be a
503 problem if you intend to base commercial development on llvm-gcc without
504 redistributing your source code.</p>
506 <p>We have no plans to change the license of LLVM. If you have questions
507 or comments about the license, please contact the <a
510 </div>
512 <!-- _______________________________________________________________________ -->
516 <p>To the best of our knowledge, LLVM does not infringe on any patents (we have
517 actually removed code from LLVM in the past that was found to infringe).
518 Having code in LLVM that infringes on patents would violate an important
519 goal of the project by making it hard or impossible to reuse the code for
520 arbitrary purposes (including commercial use).</p>
522 <p>When contributing code, we expect contributors to notify us of any potential
523 for patent-related trouble with their changes. If you own the rights to a
524 patent and would like to contribute code to LLVM that relies on it, we
525 require that you sign an agreement that allows any other user of LLVM to
526 freely use your patent. Please contact the <a
528 details.</p>
529 </div>
532 <!-- _______________________________________________________________________ -->
535 <p>With regards to the LLVM copyright and licensing, developers agree to
536 assign their copyrights to UIUC for any contribution made so that
537 the entire software base can be managed by a single copyright holder. This
538 implies that any contributions can be licensed under the license that the
539 project uses.</p>
540 </div>
542 <!-- *********************************************************************** -->
543 <hr>
544 <address>
549 Written by the
552 Last modified: $Date$
553 </address>
554 </body>
555 </html>