--disable-asm flag not being fully honored

[Chris Ballinger]

Yeah that sounds good to me. I wasn't trying to start a compiler war...

It would be nice to eventually be able to use the ARM and x86 (especially
AES-NI) assembly with clang in a way that maintains compatibility with GCC,
but I'm not exactly sure what that would entail.

Also, the --disable-asm flag I think should be renamed to --disable-mpi-asm
if that's all it does and a new option called --disable-asm actually
disable all use of assembler. Or at the very least mention in the docs that
--disable-asm doesn't disable asm and you need to manually pass -DNO_ASM in
your CFLAGS.


On Fri, Sep 20, 2013 at 7:22 AM, Jussi Kivilinna <jussi.kivilinna at iki.fi>wrote:

> On 20.09.2013 14:53, Werner Koch wrote:
> > On Fri, 20 Sep 2013 11:44, jussi.kivilinna at iki.fi said:
> >
> >> Maybe we could add Clang detection to configure.ac and then #undef
> __GNUC__ if compiler is Clang. This would be because "GCC support" in Clang
> appears to be broken.
> >
> > ./configure --disable-heinous-clang-illness
> >
> > ?
>
> Users are still going to try Clang and complain when libgcrypt does not
> work out of box with it.
>
> Maybe this should be dealt as portability issue and accept fact that there
> are broken compilers that lie about their ability to behave as GCC within
> '#ifdef __GNUC__/#endif'. So instead of checking for Clang specifically, we
> could add check for broken __GNUC__.
>
> -Jussi
>
> >
> >
> > Shalom-Salam,
> >
> >    Werner
> >
>
>
> _______________________________________________
> Gcrypt-devel mailing list
> Gcrypt-devel at gnupg.org
> http://lists.gnupg.org/mailman/listinfo/gcrypt-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: </pipermail/attachments/20130920/5ef01c10/attachment-0001.html>