Re: Status of the 'standard' and 'base' system for the amd64 port
Andreas Jochens <firstname.lastname@example.org> writes:
> Currently, the following 12 packages of the 'standard' and 'base'
> system need a patch to build and/or work on the amd64 architecture:
> aptitude : Add 'template <>' in one place
> base-files : Create 'lib64' -> 'lib' FHS 2.3 compatibility symlinks
> db4.2 : Replace '"$LD"' by '$LD' two times in 'configure'
> Add alternative Build-Depends on libgcj5-dev [amd64]
> debootstrap : Install 'libstdc++6' as 'required' on amd64
Thats from gcc-3.4. Why is that required for pure64?
> gcc-defaults : Change the defaults (for amd64 only)
Again, why? We haven't yet switched over to 3.4 and its largely
untested. I would be nice but would create a lot more broken packages
> gpm : Apply patch from BTS to fix FTBFS (not amd64 specific)
> : Remove one 'default:' label, add one 'break;' statement
> ifupdown : Add one 'break;' statement
> libsigc++-1.2 : Replace '"$LD"' -> '$LD' two times in 'configure'
> mpack : Remove wrong declarations of 'malloc' and 'getenv'
That shoulöd hit other 64bit archs too I think.
> mtr : Replace Build-Depends on 'automaken' by 'automake1.4'
> Add 'autotools-dev' to Build-Depends
> Copy config.guess and config.sub from 'autotools-dev'
> Change 'res_mkquery' to '__res_mkquery' in resolver
> library check in 'configure.in' and 'configure'
This should hit ia64 too according to glibc docs. Not sure what they
did to work around it.
> netkit-base : Add three missing 'break;' statements
> pam : Add one missing 'continue;' statement
> Bug reports with patches have been filed to the BTS for almost all of
> these issues. The patches can also be found at
> Please note that all of those patches are really short and straight
> These patches will either fix a general FTBFS which also occurs on other
> architectures or they will just change things specifically for amd64
> without having any effect on other architectures.
> If these 12 patches can somehow be applied and uploaded to the main
> archive, the 'sid' source distribution will provide a fully working
> 'standard' and 'base' system for amd64.
> For a working 'standard' and 'base' system in 'sarge',
> the amd64 port will also need the newest 'sid' version of glibc
> (glibc-2.3.2.ds1-15), which is not (yet) in 'testing'.
The maintainer said he would upload a version to t-p-u once that
works that is contents wise like sid but a lower version as needed for
> Andreas Jochens