I will try tomorrow, is have a mini-buildd that compiled for i386 and amd64
g. Marc GPG: 827C FD74 BA46 8152 A041 F3A0 7A6A 4F17 5995 A65B On Tue, 1 Sept 2026, 12:12 Simon McVittie, <[email protected]> wrote: > On Tue, 01 Sep 2026 at 12:01:23 +0200, Marc Leeman wrote: > >I opened an upstream ticket yesterday > > I saw, I've marked the bug as forwarded. > > >> >convert_fc32_to_int32: > >> >dest array 0 bad > >> > backup function : FAILED > >> > compiled function: PASSED > > This seems to be the actual failure, the others seem to be expected or > otherwise harmless. > > It looks like building orc with either -msse2 or -msse2 -mfpmath=sse > avoids this test failure. > > It's possible that -ffloat-store or -fexcess-precision=standard would > also work, but I'm not sure that it's really useful to spend time > experimenting with those, given that the Debian 13 release notes already > say that it's OK to require SSE2 now (and certainly enabling SSE2 seems > like it's more appropriate for the purpose of liborc, which is to speed > up GStreamer, if I understand correctly). > > Perhaps consider the attached? But note that I haven't been able to test > it yet, because my **amd64** build in my usual environment (qemu/kvm) is > currently failing (which is why I haven't tagged the bug + patch). > > Thanks, > smcv >
