ClangBSD build failures

Dominic Fandrey kamikaze at
Tue Apr 27 09:07:51 UTC 2010

On 27/04/2010 10:46, Roman Divacky wrote:
> I see whats going on... you have CC=cc and CXX=c++ in your share/mk/
> and the "c++" is clang thus the 

Actually I set CC and CXX in the environment. The definition in is CC?=, so it shouldn't be a problem.

> .if ${CC} == "clang" || ${CXX} == "clang++"
> MMINTRIN_CLANG= -isystem ${WORLDTMP}/usr/include/clang/1.5
> .endif
> condition does not add the -isystem thus the gcc mmintrin.h is used.
> you have to change the share/mk/ to have CC/CXX either "gcc/g++"
> or "clang/clang++"
> I have to think of some better way to test this as this proves way
> too fragile :(

Thanks a lot for your quick analysis. I already changed my test script
to gcc and g++. Though I didn't think it would have an effect, I
figured sticking to the guide would generally be a good idea.

I didn't rerun the test, though. I have to wait for the night, when
the machine is not in use.

Anyway, thank you for figuring this out so quickly. It would seriously
have confused me that it would start working for no apparent reason
and I'd probably have blamed my hardware.

> thnx for the report!

I'm working on a small clang introductory paper, and some
self-performed test seemed like the necessary salt, to give it the
right flavour. If you're interested I can notify you or the entire
list, when it's done.


A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail? 

More information about the freebsd-current mailing list