On Thu, Feb 29, 2024 at 10:24:06AM +0100, Julian Andres Klode wrote: > Control: tags 1060246 + patch > Control: tags 1060246 + pending > > Dear maintainer, > > I have prepared the following NMU, taking into consideration the > feedback to only override the ABI on the affected architectures. > > I'm not sure I want to upload it now, but as it stands, anything > build-depending on libapt-pkg-perl and other perl bindings to t64 > libraries will FTBFS. > > I uploaded it to Ubuntu but it's stuck waiting for db5.3; I believe > vorlon has done binary-uploads for db5.3 in Debian to unblock it > there though. > > Let me know what you think.
Thanks. It's clear that the perlapi names need to be bumped on the affected ports like this patch does. Just binNMU'ing perl as-is would lead to a perl that can't load the binary modules in arch:any lib*-perl packages in the archive. I'm much less clear on whether a libperl5.38t64 package will be needed. Any necessary NMUs are welcome (assuming the uploader takes the normal responsibility of course). I don't intend to upload time64 related changes myself (at least not on any kind of a short notice after waiting a month for comments.) No idea if the test failures mentioned in this bug will block things, but I'm sure the uploader will find out. That said, > +ifeq (,$(findstring $(DEB_HOST_ARCH), amd64 arm64 hurd-i386 i386 mips64el > ppc64el riscv64 s390x)) isn't this a substring match where i386 already covers hurd-i386? Guess that doesn't matter much though and none of those words match any intended port architectures. -- Niko
