🚀 go-pugleaf

In article <MUTS.93Nov13001723@compi.hobby.nl> muts@compi.hobby.nl (Peter Mutsaers) writes:
>>> On Wed, 10 Nov 1993 21:15:58 GMT, przemek@rrdbartok.nist.gov
>>> (Przemek Klosowski) said:
>
>  PK> Would Workshop and SGI cc use proprietary extensions to the
>  PK> executable format?  Does anyone know if there is a patch or
>  PK> workaround to be used with gcc?
>
>SGI uses proprietary extensions more and more, and I am beginning to

Certainly XCOFF was and is non-standard: only MIPS_based machines use
it (some others use COFF, not quite the same!).    The symbol table has
always been unique to MIPS_based machines.

With IRIX4 there was a minor change or two in the symbol table. And
ELF, with IRIX5, uses the same symbol table information in an ELF
section (named .mdebug in ELF).  There have been two more (small
changes) in the IRIX5 ELF .mdebug table in the last few weeks to
accurately model compiler-generated (as opposed to user-specified) C
typedefs and Fortran pointer variables.

We don't consider the format a secret (at least nobody ever told me
such a (IMO ridiculous) thing).  It's true we don't have good
documentation of it, which is unfortunate.  I don't think we're
publishing anything on it at present.

The basics are all still correct in the mips-stfile stuff in gcc.     
(They have to be, since nothing really significant has changed...)
I don't have time to work on gcc.     If somebody is fixing gcc and needs
information they can contact me.

Regards,
[ David B. Anderson             (415)390-4263             davea@sgi.com ]
[                      Visualize  Whirled Peas                          ]