Control: tags -1 -moreinfo +confirmed
On Fri, 2017-06-30 at 05:07 +0200, Cyril Brulebois wrote:
> gregor herrmann <gre...@debian.org> (2017-06-28):
> > On Wed, 28 Jun 2017 00:51:33 +0200, Cyril Brulebois wrote:
> > > I haven't matched this to code changes at first glance. For the sake
> > > of clarity: this relates to the Depends → Recommends update, because
> > > code was added to “apache2_invoke enmode perl” where needed?
> > Thanks for asking; this made me look at the changes again, and made me
> > realize that I made a mistake (I took only one of Axel's commits between
> > 2.5.0-4 and 2.5.0-5 but there were actually three). Sorry for that.
> > > (The second sentence makes it look like this /was/ the case already, while
> > > this seems to /become/ the case with this particular upload AFAIUI.)
> > The problem in #810655, as I understand it, is that d/control has
> > libapache2-mod-perl2 in Recommends (which is correct as embperl doesn't
> > require it) but that embperl's /etc/apache2/mods-available/zembperl.load
> > unconditionally tried to load mod_perl.
> > Axel has in a later commit removed the changes in libembperl-perl.postinst
> > again, after verifying that embperl installs without mod_perl, with mod_perl
> > installed and activated and with mod_perl installed but disabled.
> Tests are good. :)
> > So the only remaining code change is actually:
> > #v+
> > --- a/debian/zembperl.load.in
> > +++ b/debian/zembperl.load.in
> > @@ -1,6 +1,6 @@
> > # The sucky "zembperl" name is so we load after perl
> > -# Depends: perl
> > +# Recommends: perl
> > <IfModule mod_perl.c>
> > LoadModule embperl_module @ARCHLIB@/auto/Embperl/Embperl.so
> > #v-
> > I've now tentatively changed d/changelog to say
> > #v+
> > * Change hard dependency on mod_perl in zembperl.load to Recommends.
> > mod_perl is not required, and is enabled by default anyway if it is
> > installed.
> > This change matches the package dependencies and fixes an installation
> > failure when libapache2-mod-perl2 is not installed.
> > (Closes: #810655)
> > #v-
> > Does this make sense?
> I think the situation is clearer with your explanations above, and the
> changes+changelog look in sync and reasonable.
> > I'm attaching the full new debdiff, and I'm looping in Axel for a sanity
> > check.
> I won't be tagging this bug report with +confirmed right away, since we're
> awaiting for some more feedback, but the proposed changes look good to me.
Me too. Based on the subsequent responses, please go ahead.