Re: ivtools for ATrpms

On Wed, Jan 12, 2011 at 6:25 AM, Axel Thimm <[email protected]> wrote:

> On Tue, 2011-01-11 at 17:25 -0200, Paulo Cavalcanti wrote:
> > This new version should built in ATrpms. This package has the most
> > complicated configuration scheme
> > I have ever seen, and I hope you do  not mind using only one dirt
> > trick, which does not affect anything ...
>
> It would be nicer to understand why the buildroot gets stamped into the
> binary and take care of it at the build level, but let's push it that
> way. It still fails on i386 due to
>
> %ifarch %{ix86}
> pushd %{_builddir}/../BUILDROOT/
> mv %{name}-%{version}-%{release}.1 %{name}-%{version}-%{release}.i386
> popd
> %endif
>
> You cannot know what the %{buildroot} is, this package is tailored to
> only work in mock, and even there if the variables change the build
> breaks. The other fix before is also prone to buildsystem setup, e.g.
> what if a buildsystem uses a buildroot like a plain
> /var/tmp/build-<random> string?
>
>
I plan to fix it, but I still have no idea why it uses .1 and not .i386 as a
suffix for 32 bits during the build phase.
This breaks the install phase for 32 bits.

At least now I know the package builds in ATrpms as soon as I clean these
two hacks.

Thanks.

-- 
Paulo Roma Cavalcanti
LCG - UFRJ