Current FTBFS packages (was Re: [ACTION REQUIRED] Retiring packages for Fedora 21)

[Peter Jones]

On Wed, Jun 18, 2014 at 02:16:49PM -0400, Adam Jackson wrote:
> On Mon, 2014-06-09 at 14:18 -0400, Adam Jackson wrote:
> 
> > > libguestfs uses hfsplus-tools in order to provide some HFS+ filesystem
> > > features (mainly for Mac filesystems and .DMG files).  We can remove
> > > this functionality from the Fedora version, but of course it means
> > > people won't be able to perform some operations on Mac filesystems.
> > 
> > Yeah, I'd prefer if we didn't retire hfsplus-tools too.  It'd be nice if
> > this got some attention from the arm guys, I tried to force llvm to
> > default to hard-float in 3.4-8 but it doesn't seem to have been enough
> > to fix this.  I can keep poking at it but I'm assuredly not the best man
> > for the job.
> 
> Just to follow up on this:
> 
> Richard added an ExcludeArch: %{arm} in Release: 7, so that fixed the
> FTBFS in a sense.  Florian Weimer, in a comment on bug 803433, offered
> to fix the source to not use clang's blocks feature (which was the only
> reason it needed clang to build), but I beat him to it.  With that, and
> another fix for arm's stdarg implementation being (righteously) pickier
> than others, Release: 8 of hfsplus-tools builds with gcc for all arches:
> 
> http://koji.fedoraproject.org/koji/taskinfo?taskID=7055465
> 
> If I may vent for a moment, I'd like to point out exactly how spurious
> the blocks usage was (and, implicitly, troll for code review):
> 
> http://pkgs.fedoraproject.org/cgit/hfsplus-tools.git/plain/hfsplus-tools-no-blocks.patch
> 
> That's right kids, the C89 version is less code even _before_ you count
> the actual Blocks runtime.

And even more if all the structure for multiple event callbacks that
just isn't used were removed.  What's the point of all that?

-- 
        Peter