[cfe-dev] libc++ and const expr.

Richard Smith richard at metafoo.co.uk
Wed Feb 8 15:27:45 PST 2012


Yes, the C++ core working group has (provisionally) agreed to make certain
overflowing signed left shifts implementation-defined as the resolution of
DR1457. Clang implemented support for this in r150059.

On Wed, Feb 8, 2012 at 3:03 PM, Matt Johnson <johnso87 at illinois.edu> wrote:

>  I had this problem (static_assert in include/ratio) with Clang HEAD a few days ago, but it has since been resolved (r150103).  libc++ HEAD builds cleanly for me on Linux.
> -Matt
>
> On 5 Feb 2012, at 14:07, Christopher Jefferson wrote:
>
>
> On 4 Feb 2012, at 17:08, Richard Pennington wrote:
>
> >* On Saturday, February 04, 2012 04:25:07 AM Jean-Daniel Dupas wrote:*>>* Hi,*>>* *>>* Am I the only one having trouble compiling libc++ with clang TOT.*>>* For sometime now, clang refuse to compile (and use it) because of the*>>* following issue:*>>* *>>* ../include/ratio:193:19: error: static_assert expression is not an integral*>>* ../include/ratio:189:27: note: declared here*>>*    static const intmax_t max = -min;*>>* *>>* -- Jean-Daniel*>* *>* I got the same thing.*
> A number of things seem to have broken all at once. Commenting out that static_assert leads to a bug in src/locale.cpp about switching on '-1' on a switch on size_t in two places. Fixing that bug (by replacing -1 and -2 with (size_t)-1 and (size_t)-2, which is hopefully right.
>
> Fixing that leads to lining producing:
>
> ld: bad codegen, pointer diff in __ZSt17__throw_bad_allocv to global weak symbol __ZTISt9bad_alloc for architecture i386
>
> Rather than try to debug this myself, I'm tempted to leave it to more experienced people.
>
> One problem that this has shown me is that the recommended way of building libcxx (symlinking the svn directories into /usr/include and /usr/lib) means that if you svn update the library and it won't build, you are left with a totally non-functional library.
>
> Chris
>
>
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20120208/21d8d22d/attachment.html>


More information about the cfe-dev mailing list