1.. _developer_policy: 2 3===================== 4LLVM Developer Policy 5===================== 6 7.. contents:: 8 :local: 9 10Introduction 11============ 12 13This document contains the LLVM Developer Policy which defines the project's 14policy towards developers and their contributions. The intent of this policy is 15to eliminate miscommunication, rework, and confusion that might arise from the 16distributed nature of LLVM's development. By stating the policy in clear terms, 17we hope each developer can know ahead of time what to expect when making LLVM 18contributions. This policy covers all llvm.org subprojects, including Clang, 19LLDB, libc++, etc. 20 21This policy is also designed to accomplish the following objectives: 22 23#. Attract both users and developers to the LLVM project. 24 25#. Make life as simple and easy for contributors as possible. 26 27#. Keep the top of Subversion trees as stable as possible. 28 29#. Establish awareness of the project's `copyright, license, and patent 30 policies`_ with contributors to the project. 31 32This policy is aimed at frequent contributors to LLVM. People interested in 33contributing one-off patches can do so in an informal way by sending them to the 34`llvm-commits mailing list 35<http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits>`_ and engaging another 36developer to see it through the process. 37 38Developer Policies 39================== 40 41This section contains policies that pertain to frequent LLVM developers. We 42always welcome `one-off patches`_ from people who do not routinely contribute to 43LLVM, but we expect more from frequent contributors to keep the system as 44efficient as possible for everyone. Frequent LLVM contributors are expected to 45meet the following requirements in order for LLVM to maintain a high standard of 46quality. 47 48Stay Informed 49------------- 50 51Developers should stay informed by reading at least the "dev" mailing list for 52the projects you are interested in, such as `llvmdev 53<http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>`_ for LLVM, `cfe-dev 54<http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev>`_ for Clang, or `lldb-dev 55<http://lists.cs.uiuc.edu/mailman/listinfo/lldb-dev>`_ for LLDB. If you are 56doing anything more than just casual work on LLVM, it is suggested that you also 57subscribe to the "commits" mailing list for the subproject you're interested in, 58such as `llvm-commits 59<http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits>`_, `cfe-commits 60<http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits>`_, or `lldb-commits 61<http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits>`_. Reading the 62"commits" list and paying attention to changes being made by others is a good 63way to see what other people are interested in and watching the flow of the 64project as a whole. 65 66We recommend that active developers register an email account with `LLVM 67Bugzilla <http://llvm.org/bugs/>`_ and preferably subscribe to the `llvm-bugs 68<http://lists.cs.uiuc.edu/mailman/listinfo/llvmbugs>`_ email list to keep track 69of bugs and enhancements occurring in LLVM. We really appreciate people who are 70proactive at catching incoming bugs in their components and dealing with them 71promptly. 72 73.. _patch: 74.. _one-off patches: 75 76Making a Patch 77-------------- 78 79When making a patch for review, the goal is to make it as easy for the reviewer 80to read it as possible. As such, we recommend that you: 81 82#. Make your patch against the Subversion trunk, not a branch, and not an old 83 version of LLVM. This makes it easy to apply the patch. For information on 84 how to check out SVN trunk, please see the `Getting Started 85 Guide <GettingStarted.html#checkout>`_. 86 87#. Similarly, patches should be submitted soon after they are generated. Old 88 patches may not apply correctly if the underlying code changes between the 89 time the patch was created and the time it is applied. 90 91#. Patches should be made with ``svn diff``, or similar. If you use a 92 different tool, make sure it uses the ``diff -u`` format and that it 93 doesn't contain clutter which makes it hard to read. 94 95#. If you are modifying generated files, such as the top-level ``configure`` 96 script, please separate out those changes into a separate patch from the rest 97 of your changes. 98 99When sending a patch to a mailing list, it is a good idea to send it as an 100*attachment* to the message, not embedded into the text of the message. This 101ensures that your mailer will not mangle the patch when it sends it (e.g. by 102making whitespace changes or by wrapping lines). 103 104*For Thunderbird users:* Before submitting a patch, please open *Preferences > 105Advanced > General > Config Editor*, find the key 106``mail.content_disposition_type``, and set its value to ``1``. Without this 107setting, Thunderbird sends your attachment using ``Content-Disposition: inline`` 108rather than ``Content-Disposition: attachment``. Apple Mail gamely displays such 109a file inline, making it difficult to work with for reviewers using that 110program. 111 112.. _code review: 113 114Code Reviews 115------------ 116 117LLVM has a code review policy. Code review is one way to increase the quality of 118software. We generally follow these policies: 119 120#. All developers are required to have significant changes reviewed before they 121 are committed to the repository. 122 123#. Code reviews are conducted by email, usually on the llvm-commits list. 124 125#. Code can be reviewed either before it is committed or after. We expect major 126 changes to be reviewed before being committed, but smaller changes (or 127 changes where the developer owns the component) can be reviewed after commit. 128 129#. The developer responsible for a code change is also responsible for making 130 all necessary review-related changes. 131 132#. Code review can be an iterative process, which continues until the patch is 133 ready to be committed. 134 135Developers should participate in code reviews as both reviewers and 136reviewees. If someone is kind enough to review your code, you should return the 137favor for someone else. Note that anyone is welcome to review and give feedback 138on a patch, but only people with Subversion write access can approve it. 139 140Code Owners 141----------- 142 143The LLVM Project relies on two features of its process to maintain rapid 144development in addition to the high quality of its source base: the combination 145of code review plus post-commit review for trusted maintainers. Having both is 146a great way for the project to take advantage of the fact that most people do 147the right thing most of the time, and only commit patches without pre-commit 148review when they are confident they are right. 149 150The trick to this is that the project has to guarantee that all patches that are 151committed are reviewed after they go in: you don't want everyone to assume 152someone else will review it, allowing the patch to go unreviewed. To solve this 153problem, we have a notion of an 'owner' for a piece of the code. The sole 154responsibility of a code owner is to ensure that a commit to their area of the 155code is appropriately reviewed, either by themself or by someone else. The list 156of current code owners can be found in the file 157`CODE_OWNERS.TXT <http://llvm.org/viewvc/llvm-project/llvm/trunk/CODE_OWNERS.TXT?view=markup>`_ 158in the root of the LLVM source tree. 159 160Note that code ownership is completely different than reviewers: anyone can 161review a piece of code, and we welcome code review from anyone who is 162interested. Code owners are the "last line of defense" to guarantee that all 163patches that are committed are actually reviewed. 164 165Being a code owner is a somewhat unglamorous position, but it is incredibly 166important for the ongoing success of the project. Because people get busy, 167interests change, and unexpected things happen, code ownership is purely opt-in, 168and anyone can choose to resign their "title" at any time. For now, we do not 169have an official policy on how one gets elected to be a code owner. 170 171.. _include a testcase: 172 173Test Cases 174---------- 175 176Developers are required to create test cases for any bugs fixed and any new 177features added. Some tips for getting your testcase approved: 178 179* All feature and regression test cases are added to the ``llvm/test`` 180 directory. The appropriate sub-directory should be selected (see the `Testing 181 Guide <TestingGuide.html>`_ for details). 182 183* Test cases should be written in `LLVM assembly language <LangRef.html>`_ 184 unless the feature or regression being tested requires another language 185 (e.g. the bug being fixed or feature being implemented is in the llvm-gcc C++ 186 front-end, in which case it must be written in C++). 187 188* Test cases, especially for regressions, should be reduced as much as possible, 189 by `bugpoint <Bugpoint.html>`_ or manually. It is unacceptable to place an 190 entire failing program into ``llvm/test`` as this creates a *time-to-test* 191 burden on all developers. Please keep them short. 192 193Note that llvm/test and clang/test are designed for regression and small feature 194tests only. More extensive test cases (e.g., entire applications, benchmarks, 195etc) should be added to the ``llvm-test`` test suite. The llvm-test suite is 196for coverage (correctness, performance, etc) testing, not feature or regression 197testing. 198 199Quality 200------- 201 202The minimum quality standards that any change must satisfy before being 203committed to the main development branch are: 204 205#. Code must adhere to the `LLVM Coding Standards <CodingStandards.html>`_. 206 207#. Code must compile cleanly (no errors, no warnings) on at least one platform. 208 209#. Bug fixes and new features should `include a testcase`_ so we know if the 210 fix/feature ever regresses in the future. 211 212#. Code must pass the ``llvm/test`` test suite. 213 214#. The code must not cause regressions on a reasonable subset of llvm-test, 215 where "reasonable" depends on the contributor's judgement and the scope of 216 the change (more invasive changes require more testing). A reasonable subset 217 might be something like "``llvm-test/MultiSource/Benchmarks``". 218 219Additionally, the committer is responsible for addressing any problems found in 220the future that the change is responsible for. For example: 221 222* The code should compile cleanly on all supported platforms. 223 224* The changes should not cause any correctness regressions in the ``llvm-test`` 225 suite and must not cause any major performance regressions. 226 227* The change set should not cause performance or correctness regressions for the 228 LLVM tools. 229 230* The changes should not cause performance or correctness regressions in code 231 compiled by LLVM on all applicable targets. 232 233* You are expected to address any `Bugzilla bugs <http://llvm.org/bugs/>`_ that 234 result from your change. 235 236We prefer for this to be handled before submission but understand that it isn't 237possible to test all of this for every submission. Our build bots and nightly 238testing infrastructure normally finds these problems. A good rule of thumb is 239to check the nightly testers for regressions the day after your change. Build 240bots will directly email you if a group of commits that included yours caused a 241failure. You are expected to check the build bot messages to see if they are 242your fault and, if so, fix the breakage. 243 244Commits that violate these quality standards (e.g. are very broken) may be 245reverted. This is necessary when the change blocks other developers from making 246progress. The developer is welcome to re-commit the change after the problem has 247been fixed. 248 249Obtaining Commit Access 250----------------------- 251 252We grant commit access to contributors with a track record of submitting high 253quality patches. If you would like commit access, please send an email to 254`Chris <mailto:sabre@nondot.org>`_ with the following information: 255 256#. The user name you want to commit with, e.g. "hacker". 257 258#. The full name and email address you want message to llvm-commits to come 259 from, e.g. "J. Random Hacker <hacker@yoyodyne.com>". 260 261#. A "password hash" of the password you want to use, e.g. "``2ACR96qjUqsyM``". 262 Note that you don't ever tell us what your password is, you just give it to 263 us in an encrypted form. To get this, run "``htpasswd``" (a utility that 264 comes with apache) in crypt mode (often enabled with "``-d``"), or find a web 265 page that will do it for you. 266 267Once you've been granted commit access, you should be able to check out an LLVM 268tree with an SVN URL of "https://username@llvm.org/..." instead of the normal 269anonymous URL of "http://llvm.org/...". The first time you commit you'll have 270to type in your password. Note that you may get a warning from SVN about an 271untrusted key, you can ignore this. To verify that your commit access works, 272please do a test commit (e.g. change a comment or add a blank line). Your first 273commit to a repository may require the autogenerated email to be approved by a 274mailing list. This is normal, and will be done when the mailing list owner has 275time. 276 277If you have recently been granted commit access, these policies apply: 278 279#. You are granted *commit-after-approval* to all parts of LLVM. To get 280 approval, submit a `patch`_ to `llvm-commits 281 <http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits>`_. When approved 282 you may commit it yourself. 283 284#. You are allowed to commit patches without approval which you think are 285 obvious. This is clearly a subjective decision --- we simply expect you to 286 use good judgement. Examples include: fixing build breakage, reverting 287 obviously broken patches, documentation/comment changes, any other minor 288 changes. 289 290#. You are allowed to commit patches without approval to those portions of LLVM 291 that you have contributed or maintain (i.e., have been assigned 292 responsibility for), with the proviso that such commits must not break the 293 build. This is a "trust but verify" policy and commits of this nature are 294 reviewed after they are committed. 295 296#. Multiple violations of these policies or a single egregious violation may 297 cause commit access to be revoked. 298 299In any case, your changes are still subject to `code review`_ (either before or 300after they are committed, depending on the nature of the change). You are 301encouraged to review other peoples' patches as well, but you aren't required 302to. 303 304.. _discuss the change/gather consensus: 305 306Making a Major Change 307--------------------- 308 309When a developer begins a major new project with the aim of contributing it back 310to LLVM, s/he should inform the community with an email to the `llvmdev 311<http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>`_ email list, to the extent 312possible. The reason for this is to: 313 314#. keep the community informed about future changes to LLVM, 315 316#. avoid duplication of effort by preventing multiple parties working on the 317 same thing and not knowing about it, and 318 319#. ensure that any technical issues around the proposed work are discussed and 320 resolved before any significant work is done. 321 322The design of LLVM is carefully controlled to ensure that all the pieces fit 323together well and are as consistent as possible. If you plan to make a major 324change to the way LLVM works or want to add a major new extension, it is a good 325idea to get consensus with the development community before you start working on 326it. 327 328Once the design of the new feature is finalized, the work itself should be done 329as a series of `incremental changes`_, not as a long-term development branch. 330 331.. _incremental changes: 332 333Incremental Development 334----------------------- 335 336In the LLVM project, we do all significant changes as a series of incremental 337patches. We have a strong dislike for huge changes or long-term development 338branches. Long-term development branches have a number of drawbacks: 339 340#. Branches must have mainline merged into them periodically. If the branch 341 development and mainline development occur in the same pieces of code, 342 resolving merge conflicts can take a lot of time. 343 344#. Other people in the community tend to ignore work on branches. 345 346#. Huge changes (produced when a branch is merged back onto mainline) are 347 extremely difficult to `code review`_. 348 349#. Branches are not routinely tested by our nightly tester infrastructure. 350 351#. Changes developed as monolithic large changes often don't work until the 352 entire set of changes is done. Breaking it down into a set of smaller 353 changes increases the odds that any of the work will be committed to the main 354 repository. 355 356To address these problems, LLVM uses an incremental development style and we 357require contributors to follow this practice when making a large/invasive 358change. Some tips: 359 360* Large/invasive changes usually have a number of secondary changes that are 361 required before the big change can be made (e.g. API cleanup, etc). These 362 sorts of changes can often be done before the major change is done, 363 independently of that work. 364 365* The remaining inter-related work should be decomposed into unrelated sets of 366 changes if possible. Once this is done, define the first increment and get 367 consensus on what the end goal of the change is. 368 369* Each change in the set can be stand alone (e.g. to fix a bug), or part of a 370 planned series of changes that works towards the development goal. 371 372* Each change should be kept as small as possible. This simplifies your work 373 (into a logical progression), simplifies code review and reduces the chance 374 that you will get negative feedback on the change. Small increments also 375 facilitate the maintenance of a high quality code base. 376 377* Often, an independent precursor to a big change is to add a new API and slowly 378 migrate clients to use the new API. Each change to use the new API is often 379 "obvious" and can be committed without review. Once the new API is in place 380 and used, it is much easier to replace the underlying implementation of the 381 API. This implementation change is logically separate from the API 382 change. 383 384If you are interested in making a large change, and this scares you, please make 385sure to first `discuss the change/gather consensus`_ then ask about the best way 386to go about making the change. 387 388Attribution of Changes 389---------------------- 390 391We believe in correct attribution of contributions to their contributors. 392However, we do not want the source code to be littered with random attributions 393"this code written by J. Random Hacker" (this is noisy and distracting). In 394practice, the revision control system keeps a perfect history of who changed 395what, and the CREDITS.txt file describes higher-level contributions. If you 396commit a patch for someone else, please say "patch contributed by J. Random 397Hacker!" in the commit message. 398 399Overall, please do not add contributor names to the source code. 400 401.. _copyright, license, and patent policies: 402 403Copyright, License, and Patents 404=============================== 405 406.. note:: 407 408 This section deals with legal matters but does not provide legal advice. We 409 are not lawyers --- please seek legal counsel from an attorney. 410 411This section addresses the issues of copyright, license and patents for the LLVM 412project. The copyright for the code is held by the individual contributors of 413the code and the terms of its license to LLVM users and developers is the 414`University of Illinois/NCSA Open Source License 415<http://www.opensource.org/licenses/UoI-NCSA.php>`_ (with portions dual licensed 416under the `MIT License <http://www.opensource.org/licenses/mit-license.php>`_, 417see below). As contributor to the LLVM project, you agree to allow any 418contributions to the project to licensed under these terms. 419 420Copyright 421--------- 422 423The LLVM project does not require copyright assignments, which means that the 424copyright for the code in the project is held by its respective contributors who 425have each agreed to release their contributed code under the terms of the `LLVM 426License`_. 427 428An implication of this is that the LLVM license is unlikely to ever change: 429changing it would require tracking down all the contributors to LLVM and getting 430them to agree that a license change is acceptable for their contribution. Since 431there are no plans to change the license, this is not a cause for concern. 432 433As a contributor to the project, this means that you (or your company) retain 434ownership of the code you contribute, that it cannot be used in a way that 435contradicts the license (which is a liberal BSD-style license), and that the 436license for your contributions won't change without your approval in the 437future. 438 439.. _LLVM License: 440 441License 442------- 443 444We intend to keep LLVM perpetually open source and to use a liberal open source 445license. **As a contributor to the project, you agree that any contributions be 446licensed under the terms of the corresponding subproject.** All of the code in 447LLVM is available under the `University of Illinois/NCSA Open Source License 448<http://www.opensource.org/licenses/UoI-NCSA.php>`_, which boils down to 449this: 450 451* You can freely distribute LLVM. 452* You must retain the copyright notice if you redistribute LLVM. 453* Binaries derived from LLVM must reproduce the copyright notice (e.g. in an 454 included readme file). 455* You can't use our names to promote your LLVM derived products. 456* There's no warranty on LLVM at all. 457 458We believe this fosters the widest adoption of LLVM because it **allows 459commercial products to be derived from LLVM** with few restrictions and without 460a requirement for making any derived works also open source (i.e. LLVM's 461license is not a "copyleft" license like the GPL). We suggest that you read the 462`License <http://www.opensource.org/licenses/UoI-NCSA.php>`_ if further 463clarification is needed. 464 465In addition to the UIUC license, the runtime library components of LLVM 466(**compiler_rt, libc++, and libclc**) are also licensed under the `MIT License 467<http://www.opensource.org/licenses/mit-license.php>`_, which does not contain 468the binary redistribution clause. As a user of these runtime libraries, it 469means that you can choose to use the code under either license (and thus don't 470need the binary redistribution clause), and as a contributor to the code that 471you agree that any contributions to these libraries be licensed under both 472licenses. We feel that this is important for runtime libraries, because they 473are implicitly linked into applications and therefore should not subject those 474applications to the binary redistribution clause. This also means that it is ok 475to move code from (e.g.) libc++ to the LLVM core without concern, but that code 476cannot be moved from the LLVM core to libc++ without the copyright owner's 477permission. 478 479Note that the LLVM Project does distribute llvm-gcc and dragonegg, **which are 480GPL.** This means that anything "linked" into llvm-gcc must itself be compatible 481with the GPL, and must be releasable under the terms of the GPL. This implies 482that **any code linked into llvm-gcc and distributed to others may be subject to 483the viral aspects of the GPL** (for example, a proprietary code generator linked 484into llvm-gcc must be made available under the GPL). This is not a problem for 485code already distributed under a more liberal license (like the UIUC license), 486and GPL-containing subprojects are kept in separate SVN repositories whose 487LICENSE.txt files specifically indicate that they contain GPL code. 488 489We have no plans to change the license of LLVM. If you have questions or 490comments about the license, please contact the `LLVM Developer's Mailing 491List <mailto:llvmdev@cs.uiuc.edu>`_. 492 493Patents 494------- 495 496To the best of our knowledge, LLVM does not infringe on any patents (we have 497actually removed code from LLVM in the past that was found to infringe). Having 498code in LLVM that infringes on patents would violate an important goal of the 499project by making it hard or impossible to reuse the code for arbitrary purposes 500(including commercial use). 501 502When contributing code, we expect contributors to notify us of any potential for 503patent-related trouble with their changes (including from third parties). If 504you or your employer own the rights to a patent and would like to contribute 505code to LLVM that relies on it, we require that the copyright owner sign an 506agreement that allows any other user of LLVM to freely use your patent. Please 507contact the `oversight group <mailto:llvm-oversight@cs.uiuc.edu>`_ for more 508details. 509