Nico Golde <[EMAIL PROTECTED]> writes:
> Hi,
> does anyone knows a solution to let packages FTBFS on
> buildd's which architecure are not supported by the
> software?
> regards nico
>
> --
> Nico Golde - [EMAIL PROTECTED] | GPG: 1024D/73647CFF
> http://www.ngolde.de | http://www.muttng.org | http://grml.org
> VIM has two modes - the one in which it beeps
> and the one in which it doesn't -- encrypted mail preferred
Some things should be noted:
1. the Archtecture line of the source is relevant
- sources with only arch dependend packages only list relevant archs
Sbuild checks for this and skips bad packages.
- sources with also arch independend packages are arch any
This gives a false positve with sbuild and the package builds
often up until dpkg-gencontrol fails to find a matching arch.
It would be nice to get the Architecture field to list e.g. "all
i386 x86_64" in those cases instead of "any".
2. Build-Depends may not be on (purely) virtual packages
A "Build-Depends: i386 | x86_64" does not work. Instead the
following can be used:
Build-Depends: type-handling
Build-Conflicts: alpha arm hppa darwin-i386 freebsd-i386 hurd-i386
kfreebsd-i386 knetbsd-i386 netbsd-i386 openbsd-i386 ia64 m68k mips mipsel
darwin-powerpc powerpc s390 s390x sh3 sh3eb sh4 sh4eb sparc sparc64
The right Build-Conflicts can (should) be generate with e.g.
type-handling -n i386,x86_64 linux
3. debian/rules can check the arch
Some package should build on archs but don't, e.g. kernel-patch-arch
packages that are arch:all still only build on their own arch none
the less. Others should fail but instead do serious harm to buildds
if tried on the wrong arch, e.g. kernel-patch packages going into
an endless loop.
4. packages-arch-specific can remove packages from being autobuild on
certain archs at all
This, unlike 2 or 3 above, actualy prevents the buildd and buildd
admins from wasting any time on your package. It does not help
developers that try to build your package though.
MfG
Goswin
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
