Re: Additional link flags for HP aCC and SGI CC

[Gary V. Vaughan]

Salut Ludovic,

Ludovic Courtès wrote:
> Yesterday, 22 hours, 53 minutes, 20 seconds ago, Gary V. Vaughan wrote:
> 
>>If you'd like to help out with this (-Wall might be a good place to start),
>>and you are in a position to assign copyright of your patches to the FSF,
>>I for one would be delighted to help you out.
> 
> 
> I started to look at `libtool.m4' to get an idea of how things work by
> looking eg. at how `lt_prog_compiler_static_*' are determined.  First, I
> must say that I am very impressed by the number of supported systems.

If you haven't already, you might want to start with CVS head, where a lot of
refactoring makes the code much easier to follow.

> However, I am surprised by some of the assumptions that are made
> regarding the compiler used on a given OS platform while, for instance,
> GCC may be available on most of them.  Instead, I used to determine the
> compiler being used by checking for particular preprocessor macros
> (eg. `__GNUC__', `__DECCXX', etc.).  Is there any reason for not doing
> this?  Maybe such macros are not always documented or may be changing
> over time?

Due to vendors changing things between releases.

For each language (tag in libtool terminology), we try to support the gnu
compiler collection and the vendor compilers for each supported architecture.
The assumption is that if we don't detect gcc (g++, gcj etc.), then it is
the vendor's compiler we are running.  We are also careful about whether
gcc uses gnu ld, or the vendor linker.  This means that in many cases, libtool
well continue to work when the vendor revs their OS and/or compiler suite.

HTH,
        Gary.
-- 
Gary V. Vaughan      ())_.  address@hidden,gnu.org}
Research Scientist   ( '/   http://tkd.kicks-ass.net
GNU Hacker           / )=   http://www.gnu.org/software/libtool
Technical Author   `(_~)_   http://sources.redhat.com/autobook

signature.asc
Description: OpenPGP digital signature