Tom Hukins sent me a very nice note off-list thanking me for assembling a list of bug fixes for possible backporting. I'm copying it back to the list to make a bigger point about perldelta hygiene: > Thank you very much for putting together this useful list of commits > to consider for 5.18.1. I suspect it's largely thankless, tedious > work, but having as many useful bug fixes in the release will make the > release much more useful. Actually, it *wasn't* tedious. It took about 10 minutes. I copied the "selected bugs" sections from the perl5191delta and perl5192delta into my editor, sorted them into categories that fit perlpolicy and deleted the rest. The *hard* work was already done, by the release managers and committers who wrote up perldelta entries for fixed bugs and closed tickets. In an ideal world, every commit (or commit series) that closes a ticket should have a corresponding perldelta entry that summarizes the issue and fix. Ideally, that would also be the first paragraph or so of the text in the commit message, even if the commit message goes into much greater detail after that. Failing an entry in perldelta, commit messages should at least have enough description in them to be used to create a perldelta entry later. That means "fixed RT#123456" is a very bad commit message. A good commit message should make clear the issue *and* resolution. Therefore: * If you are writing a patch: please provide a perldelta entry * If you are applying a patch: please ask the sender to patch perldelta if they did not and please check the quality of the commit message before applying If we have can maintain good hygiene around commits and perldeltas, then the work of identifying bug fixes to backport becomes much easier. David -- David Golden <xdg@xdg.me> Take back your inbox! → http://www.bunchmail.com/ Twitter/IRC: @xdgThread Next