Ok, everything working fine after deleting the previous cmake build files.

Am Donnerstag, 20. November 2014 23:16:50 UTC+1 schrieb Floh:
>
> Wait, this could be cached cmake build files. I DID have valgrind 
> installed recently and removed it when cleaning up my brew installation. 
> I'm currently testing after manually deleting the 'build_incoming_64' 
> directory under fastcomp. Will take a little while...
>
> Am Donnerstag, 20. November 2014 23:07:38 UTC+1 schrieb Floh:
>>
>> I'm getting a compile error with the emsdk with './emsdk install 
>> sdk-incoming-64bit' 
>> that <valgrind/valgrind.h> could not be found, currently investigating. 
>> There's an #ifdef HAVE_VALGRIND_VALGRIND_H in Support/Valgrind.cpp. 
>> Investigating...
>>
>> [  6%] Building CXX object 
>> lib/Support/CMakeFiles/LLVMSupport.dir/Valgrind.cpp.o
>> /Users/floh/projects/oryol/sdks/osx/emsdk_portable/clang/fastcomp/src/lib/Support/Valgrind.cpp:20:10:
>>  
>> fatal error: 
>>       'valgrind/valgrind.h' file not found
>> #include <valgrind/valgrind.h>
>>          ^
>> 1 error generated.
>> make[2]: *** [lib/Support/CMakeFiles/LLVMSupport.dir/Valgrind.cpp.o] 
>> Error 1
>>
>> Am Donnerstag, 20. November 2014 22:06:30 UTC+1 schrieb Alon Zakai:
>>>
>>> LLVM 3.4 from the pnacl tree has been merged into our incoming branches, 
>>> and the version number bumped to 1.27.1. See 
>>> https://github.com/kripken/emscripten-fastcomp/issues/51 for more 
>>> details. aidanhs did most of the hard work here - thanks!
>>>
>>> This is after a very long period of not updating LLVM or clang, mainly 
>>> due to less of a need (those updates are less significant than the 2.8-3.0 
>>> days), also we wanted to let fastcomp stabilize without LLVM changes 
>>> happening, and finally just a matter of resources. But it's a good idea to 
>>> stay as close to upstream as possible.
>>>
>>> Our test suite passes (locally - bots are starting up now) and fuzzing 
>>> can't find any issues, but this is a large update, so please test carefully.
>>>
>>> Otherwise there shouldn't be major changes because of this. The update 
>>> fixes various bugs in LLVM, but doesn't bring major features as far as I 
>>> know, and no significant perf changes. Only thing I can think of offhand is 
>>> there is the optional 'mergefunc' pass which I hear is useful at reducing 
>>> code size, which works in 3.4 but didn't earlier.
>>>
>>> Note that LLVM stable is 3.5, and we are merging 3.4 here. That means we 
>>> are still 1 version behind (6 months). After 3.4 has been stable for a 
>>> while, we can start to consider when to move to 3.5. This is a larger 
>>> change as I hear they modified some core internal APIs, and also started to 
>>> require that the toolchain that builds LLVM support c++11. So we'll need to 
>>> be careful about when we do that.
>>>
>>> - Alon
>>>
>>>

-- 
You received this message because you are subscribed to the Google Groups 
"emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to