Re: FWD: Accepted debhelper 6.0.10 (source all)
On Thu, Mar 27, 2008 at 07:11:59PM +0100, Julian Mehnle wrote:
> Updating hundreds of packages to require debhelper 6 isn't any less work
> than updating hundreds of packages to not perform an unconditional rmdir,
> plus the former hinders backportability.
These are not alternative actions. An 'unconditional rmdir' is still
going to make the package fail to build on Perl 5.10.0, with or without
the new dh_perl functionality.
> So while I generally agree with your statement, in this particular case it
> just isn't a good rationale for requiring debhelper 6 in all the packages
> affected.
In this particular case, I think an "ideal" package should neither do
the rmdir nor build-depend on debhelper (>= 6.0.10). When built on a
system with a new enough debhelper or a new enough ExtUtils::Install
(ie. Perl 5.10), the empty directory doesn't end up in the .deb. Otherwise
it does, but that doesn't really hurt anyone. Remember, this is just a
minor cleanup issue, so keep it simple.
So next time a package with a 'conditional rmdir' is updated, I think it's
best just to remove the rmdir line from debian/rules. (There's no need
to do this at once in all the hundreds of packages this time, of course.)
Cheers,
--
Niko Tyni ntyni@debian.org
Reply to: