From: brian ford Date: 2009-08-27T01:42:39+09:00 Subject: [ruby-core:25142] Re: Patch writer's guide to submit Hi Matz, These guidelines sound very reasonable. Thank you for providing them. Can I make one suggestion: please, please make test cases mandatory for all patches. Unless it is pure refactoring, which I suggest would be quite rare, the patch *must* change something. So the author must have had some Ruby code that did one thing, then after the patch, the Ruby code does something else. I believe making the test requirement mandatory is not too much to ask of patch authors. Thanks, Brian On Wed, Aug 26, 2009 at 6:06 AM, Yukihiro Matsumoto wrote: > Hi, > > Sometimes we laze core developers have procrastinated to review > submitted patches, been late to merge patchs, and frustrated > submitters. �That's unfortunate for both side. �Here's the guideline > to avoid such miscommunication. > > �* one modification per a patch > > � This is the biggest issue for most deferred patches. �When you > � submit a patch that fix multiple bugs (and add features) at once, > � we have to separate them before applying it. �It is kinda hard task > � for us, busy developers, so this kind of patches tend to be > � deferred. �No big patches please. > > �* more description > > � Sometimes mere patches do not describe problems that fix. �Better > � description (a problem that fix, precondition, platform, etc.) > � would help a patch to be merged earlier. > > �* diff to the latest > > � Your problem might have been fixed in the latest. �Or the code > � might be totally different. �Before submitting a patch, try fetch > � the latest (trunk for 1.9, ruby_1_8 for 1.8) from Subversion, > � please. > > �* diff -u > > � We perefer diff -u style unified diff patches to diff -c or any > � other style of patches. �They are far easier to review. �Do not > � send modifed files, we don't want to make diff by ourselves. > > �* (optional) test cases > > � a patch to test cases (preferably in a patch to test/*/test_*.rb) > � would help us understand the patch and your intention. > > We might move to git style push/pull in the future. �But until then, > the above guideline would help you to avoid your frustration. > > Thank you. > > > � � � � � � � � � � � � � � � � � � � � � � � � � � � �matz. > >