Hi! On Tue, 2017-03-28 at 20:52:51 +0200, Martin Quinson wrote: > On Tue, Mar 28, 2017 at 06:39:09PM +0200, Guillem Jover wrote: > > On Tue, 2017-03-28 at 09:22:04 +0200, Martin Quinson wrote: > > > as you can see, I fixed this bug without integrating your patch. This > > > is because po4a is only a parser, not a generator of such files. There > > > is nothing we can do to fix the debian changelog files. Instead, we > > > must deal with all variants of such files. Your change would have make > > > po4a applicable to even less existing changelogs... > > > > Hmm, I'm not sure I see the point of this alternative fix. As stated > > in the referred mail thread and in the commit messages, dpkg has never > > accepted correct timezones, so there should be no such changelogs in > > the wild. I'm very skeptical there are changelogs with a one-character > > timezone, but if there are, dpkg and various other tools, have stopped > > accepting them now anyway. So if your concern is that there might be > > such broken changelog (with a one-character timezone) and you still > > want to accept them (I'm not sure why? as they'd need to be fixed > > anyway to be usable :), the better solution would be to just do > > nothing and leave the broken regex as is. > > You are right, I messed up my regexp. I meant to accept timezones > given with any amount of chars. Now this part reads: \([^\\\(\)]+\)
Oh, I didn't actually check your change, only that you tried to fix the regex instead of removing that part! My comment on the broken regex was referring to the original one. > Does it seem better to you? Not really, :) because it seems a bit pointless, as dpkg has never and will never accept such correct timezones, so they will never appear in the wild. And it now does not even accept the broken one-character timezones either. :) In any case, I'm not really bothered either way. The bug was just a heads-up for other code-base maintainers, so that they could drop dead-code which has never worked. And to avoid further spreading it in case others derive new code from such code-bases. Thanks, Guillem
