The impact of multiarch on upstream Python

[Colin Watson]

On Thu, Apr 21, 2011 at 10:10:04PM -0400, Barry Warsaw wrote:
> On Apr 20, 2011, at 10:19 PM, Matthias Klose wrote:
> >> * Consideration for backward compatibility so that Python's build would
> >>    continue to work without change.
> >
> >that (having symlinks for the files in the old locations) would defeat the
> >benefit of multiarch.
> 
> Wouldn't it give packages and projects a reasonable transition period to adapt
> to the changes?  Did multiarch have to be an all-or-nothing change?

If you install those symlinks, then the same package from multiple
architectures can't be coinstallable - libfoo1:i386 would ship
/usr/lib/libfoo.so.1 -> i386-linux-gnu/libfoo.so.1, while libfoo1:amd64
would ship /usr/lib/libfoo.so.1 -> x86_64-linux-gnu/libfoo.so.1.  You
can't have both of those on the system at the same time.

The entire point of multiarch is to allow the same package from multiple
architectures to be coinstallable - so yes, this part of it did
unfortunately have to be all-or-nothing.

-- 
Colin Watson                                       [cjwatson at ubuntu.com]