On Tue, 15 Apr 2025 at 16:52:56 +0200, Helmut Grohne wrote:
On Tue, Apr 15, 2025 at 01:47:54PM +0100, Simon McVittie wrote:I think a regression for amd64/i386 co-installation would have a considerably larger practical negative impact on Debian usersThis reasoning convinces me. As it stands, I only see solutions to this problem that are inappropriate for trixie.
I think the "option 2" that I proposed is entirely feasible for trixie,
actually. I'm testing an implementation now.
What I said before is that I don't think it's feasible to send
it upstream for trixie, but it's fairly straightforward to do as a
downstream adjustment (in debian/rules rather than in the upstream code),
so inability to upstream it is not necessarily a blocker IMO. Within
Debian, we can make simplifying assumptions like "we are installing into
a $libdir that the dynamic linker will search by default" which would
not necessarily be considered valid upstream.
smcv