On 07/12/13 4:41 AM, Landry Breuil wrote:
On Fri, Dec 06, 2013 at 03:26:55PM -0500, Brad Smith wrote:
On 05/12/13 3:35 AM, Brad Smith wrote:
Here is a work in progress update to LLVM 3.4. This has only been lightly
tested on amd64. I need this build tested on i386, powerpc, sparc64 and
mips64 to start off. If you have such an arch please try building it as
soon as possible as I'd like to report any issues upstream if there are
any. Any runtime testing feedback would also be appreciated.

Still looking for build tests for sparc64 and mips64.

It builds and packages on sparc64:

$make test
===>  Regression tests for llvm-3.4rc2
ninja: error: unknown target 'RUNTESTFLAGS='

I have a diff pending with Stuart to fix the regression tests. That
was broken after the switch to CMake.

And it still fails to pass firefox's configure script:

Mozilla's autoconf infrastructure for whatever reason is not detecting that that is Clang and running autoconf checks for GCC on Clang which it should not be. I would need to see the full autoconf run up to that
point to see if we can spot where things go wrong. The autoconf script
from the bits I looked at seems to be able to determine GCC vs Clang
but I didn't come across the actual tests to see how it does so.

checking for gcc pr39608... yes
This compiler would fail to build firefox, please upgrade.
------ config.log ------
int main(void) {
   jsop_setelem(0, 47);
}

configure:11044: checking for gcc pr39608
configure:11074: c++ -c -O2 -pipe -fno-exceptions -fno-strict-aliasing
-fno-rtti -ffunction-sections -fdata-sections -fno-exceptions
-fno-math-errno -std=gnu++0x -fno-tree-vrp -Qunused-arguments
conftest.C 1>&5
clang-3.4: error: unknown argument: '-fno-tree-vrp'
configure: failed program was:
#line 11056 "configure"
#include "confdefs.h"

typedef void (*FuncType)();
template<FuncType Impl>
void f();
template<typename T> class C {
   typedef C<T> ThisC;
   template<int g()>
   static void h() {
     f<ThisC::h<g> >();
   }
};

int main() {
true
; return 0; }


Landry




--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

Reply via email to