Re: [Pkg-ia32-libs-maintainers] Bug#679671: ia32-libs-i386: depends on removed package libdb4.8
On 2012-07-18 13:34, Goswin von Brederlow wrote:
> [...]
>
> There are still 2 problems remaining: lestiff2 (patch in BTS) and cyrus-sasl
> (needs a sourcefull upload to fix binNMU / multiarch problem).
>
A patched version of lesstif2 appears to have been uploaded. To my
knowledge it caused mtink to FTBFS, which has been fixed (see #683316).
There have also been noticed issues with gromacs and cmucl, but these
appear to be unrelated to the multi-arch conversion[1]. I have CC'ed
Peter Samuelson, who have been conducting most of the lesstif2 rdep
rebuilds.
It is my understanding that lesstif2 is the last M-A conversion needed
for ia32-libs. Assuming the changes to lesstif2 and mtink are
reasonable, cyrus-sasl just needs a "no-change" upload and ia32-libs is
good to go.
> [...]
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=677762
>
> The number of issues have gone down but not far enough yet. Most
> critical and trivial there is:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=650777
>
> That one prevents ia32-libs-gtk to be installable with a dependency on
> libgnomecanvas and 3rd party binaries still depend on that.
>
> [...]
It seems like most of those bugs are libraries needing converted to M-A
(i.e. usr/lib -> usr/lib/$MADIR) and #650777 is the "exception".
What do we "lose" if only the "trivial" bugs are fixed? By trivial I
mean only patches that add "Multi-Arch: $val" fields like #650777 (so no
moving of libraries from usr/lib to usr/lib/$MADIR).
~Niels
[1] The gromacs issue appears to be #680825, which was filed before the
M-A conversion. cmulc allegedly to have a broken "build-arch" setup;
not sure if a bug has been reported for that yet.
Reply to: