PowerPC Packages

[Mark Millard]

> 10.0 on all platforms used 3.3. CURRENT uses 3.4. One interesting 
> question is that we currently build packages for a whole branch (10.X, 
> say) on 10.0. If 10.0 has a crummy compiler, that model may not be the best.
> -Nathan

Sort of like powerpc/powerpc64 being 2nd tier for FreeBSD vs. i386 and amd64 being 1st tier: clang's powerpc/powerpc64 support probably lags in time some relative to the status of its i386/amd64 support. So with 10.0-RELEASE for FreeBSD for i386/amd64 being based on clang 3.3 there was probably work on clang to make 3.3 sufficient for i386/amd64 during 10.0-RELEASE's development period. FreeBSD was probably a major test case for clang's i386/amd64 support.

Since powerpc/powerpc64 FreeBSD was not switched over to clang at the same time it would not be surprising for it to take a later clang version to get the same sort of results. (Again FreeBSD would probably be a major test case, just for a different target processor family.)


Side note relative to FreeBSD 9.X:

I'm pretty sure that when I experimented with installing FreeBSD-9.2-STABLE-powerpc-powerpc64-20140426-r264969-dvd1.iso and tried "clang --version" it reported being 3.4, not 3.3. (But I'm not currently where I can check and I've not yet investigated how to figure such out from just looking around the FreeBSD's web pages.) So it appears there has been some experimentation with clang 3.4 for power/powerpc64 outside 11.0-CURRENT. I expect that FreeBSD-9.3-BETA3-powerpc-disc1.iso and FreeBSD-9.3-BETA3-powerpc-powerpc64-disc1.iso also have clang 3.4.

But FreeBSD 9.X probably is more free to update clang because it was/is not intended to ever be used as the compiler for any FreeBSD 9.X of itself. That could mean that later 9.X's are for a time better as an environment for some types of experimentation with clang on powerpc/powerpc64 machines.

===
Mark Millard
markmi at dsl-only.net