https://blogs.gentoo.org/ago/2020/07/04/gentoo-tinderbox/ Issue: dev-libs/imath-3.1.4-r3 fails to compile on stable x86. Discovered on: x86 (internal ref: x86_at)
Created attachment 783056 [details] build.log build log and emerge --info
It's like in bug #840580. The build works with -march=i686 and fails with -march=native. Additionally, the 3.1.4-r4 and 3.1.5 build with either setting. Stabilization for x86 of 3.1.5 has stalled for some time[1]. Maybe we could stick to not using native arch flags inside a container of some sorts, when actually runnig on a native 64bit arch? At least until it's sorted out, whether this is some issue with gcc or if it's by design. [1] bug #839582
Does this still happen with current imath versions?
Seems to be solved in current v3.1.7 (see bug #905800).
(In reply to Bernd from comment #2) > It's like in bug #840580. The build works with -march=i686 and fails with > -march=native. Additionally, the 3.1.4-r4 and 3.1.5 build with either > setting. Stabilization for x86 of 3.1.5 has stalled for some time[1]. > > Maybe we could stick to not using native arch flags inside a container of > some sorts, when actually runnig on a native 64bit arch? At least until it's > sorted out, whether this is some issue with gcc or if it's by design. > > [1] bug #839582 fwiw I've started doing just -march=i686/default now as it's not worth these bugs. Similar settings would be welcome but I wouldn't recommend native. They're real bugs but ultimately not very important at all.