On Tue, Nov 23, 2004 at 12:33:29PM +0000, Steve Hay wrote: > Current practice, as stated by Nicholas earlier in this thread, is for > minor patches to be applied to blead (in the spirit of point #2 of the > Contract document), but only the latest CPAN version is actually > distributed, i.e. integrated to maint (in the spirit of point #3).
This does not appear to be the case. perl-maint contains patches against both MakeMaker 6.17 and Test::More 0.47. Of course, for most of this I mumbled an "ok" when it was done. > Hence I was suggesting p5p people should ideally have co-maintainership > of these modules to keep them in sync. Clearly each author would have > to explicitly agree to this, but if they did then it could be better for > all: New Perl releases would contain all the latest bug fixes instead > of having to leave them out just because there is no new CPAN version, > and module authors would have some of their bugs fixed for them, freeing > up their precious spare time for enhancements and other pet ideas of > their own that they'd rather spend their time on. I disagree with this. In the past I've had to reverse several "bug fixes" applied by p5p to the modules I maintain. The most recent was an attempt to fix Test::More::is_deeply() that broke something else. Fortunately I caught it before it was released. p5p has to be very, very careful about patching other's modules. You may wind up breaking backwards compat or adding a feature the maintainer does not wish to maintain. p5p should patch CPAN modules only in very clear cases (such as portability issues) or very critical cases. In all cases they should endevour to get the patch back to the author. If the author is not responding then p5p can exert more control and possibly begin the process of getting an abandoned module reassigned. But I do not agree that "all the latest bug fixes" should be applied. -- Michael G Schwern [EMAIL PROTECTED] http://www.pobox.com/~schwern/ Any sufficiently encapsulated hack is no longer a hack.
