Bug#638614: FTBFS on ia64 (internal compiler error)
So, um, just asking here: if a package fails to build on one particular
architecture due to a bug in gcc, is that a reason to remove it from
testing
If a few minor packages are blocking a transition for any reason then
it is very likely they will be removed from testing. This isn't exactly
ideal but running testing is a game of compromises.
If you want to see openmsx back in testing the out of date package on
ia64 needs to be dealt with somehow. There are basically four ways of
doing this in descending order of preference.
1: get gcc fixed.
2: if the ICE is optimisation related reduce the optimisation level
until it goes away.
3: use an older gcc version. Debian currently offers 4.4, 4.5 and 4.6
4: request removal of the out of date packages on ia64 and downgrade
this bug to "important".
1 is the best long term soloution and a gcc bug report should be
prepared and filed as a first step but based on past experiance I
would not expect a rapid response from the gcc guys.
2 is an acceptable soloution but obviously isn't ideal.
3 is acceptable as an interim soloution but is not a long term
answer because older gcc versions are eventually removed from
debian.
4 is a last resort option and should only be used after reasonable
attempts at 1-3 have failed (per the rc policy "Packages must
be supported on as many architectures as is reasonably possible.")
If you need access to an ia64 box for testing I would suggest
asking on debian-ia64@lists.debian.org
P.S. I'm just taking flyby looks at rc bugs, I have no relationship to
or interest in this particular package.
Reply to: