On Sat, Jan 14, 2017 at 10:31:34AM +0000, Eric Wong wrote: > Jeff King wrote: > > Just as a devil's advocate, why do we care about warnings in third-party > > modules? Or more specifically, why do _users_ who are running Git care > > about them? We cannot fix them in Git. A user may report the error to > > the module author, but the module author may not be responsive, or even > > may not be inclined to fix the problem (because they have a particular > > opinion on that warning). > > Every user is a potential developer(*). And I do feel > we (git developers) should be at least somewhat responsible > for helping maintain and fix the projects we depend on; > or moving to alternatives if we can't fix them. > > There is a chance a newly-introduced warning in a 3rd-party > module points to a real problem with the way git uses it, too. > Having that warning would help us fix or workaround the bug > (either in git or the module). > > I doubt any module author would be unresponsive to having a > localized "no warnings" for special cases. AFAIK, "-w" is > widespread amongst Perl users (unlike Ruby in my experience). My experience is that using strict and warnings is so common in Perl code that absent a compelling documented reason, most Perl developers would consider it a bug not to use them. Consequently, using -w, while a good practice, is unlikely to have any practical effect on external modules. What is more likely to occur is that as newer versions of Perl come out, we'll get warnings about questionable constructs that Perl, in its extensive flexibility, has permitted for a long time, but should really be fixed. Most distributors of Git will have fixed any affected third-party modules as part of the Perl upgrade, > I'm not saying we blindly start using '-w' everywhere today. > But we may at least try it and see if it introduces new > warnings, first, and only enable '-w' when it it looks quiet > (and perhaps start working with module authors to fix warnings > if not). > > As a user, I'd rather have some indication of where something > might be wrong, than no warning at all when something does go > wrong. I'm working on patches for USE_ASCIIDOCTOR=Yes and I found at least two bugs by enabling warnings. -- brian m. carlson / brian with sandals: Houston, Texas, US +1 832 623 2791 | https://www.crustytoothpaste.net/~bmc | My opinion only OpenPGP: https://keybase.io/bk2204