f77 on sco5 fails on GOTOFF references

[Joao Cardoso]

Robert Lipe wrote:
> 
> > However, here's a hacky patch that works for the test case
> >
> >       subroutine foo(a)
> >       double precision a
> >       a = dlog10(2.0d0)
> >       return
> >       end
> >
> > I've no idea if a similar change is required elsewhere as well.
> 
> Richard,  your patch does indeed stop the assembler from hurling
> on the fortran code that was sent to me that started this thread.
> Since that code was meant to be a library fragment and I don't do
> fortran, I have no way of testing that it actually executes correctly
> for that specific case.

Hi,

I might be of some help here:

I have applied the patch (and also sco5.h patch) to egcs-1.0 and rebuild
octave (a Matlab alike) that is the package from where the original fortran
file (that initiated this thread) comes from.

All 377 octave fortran files compiled OK, and octave's "make check" only 
fail on 2 of a total of 1401 tests, so I can assume that probably 
(almost) all is OK. The two failed tests have nothing to do with
Fortran, although they did not fail if compiled with gcc-2.7.2.3/g77-0.5.21.

Good news is that I can now build working shared libraries for octave,
which I was never been able to do under gcc (with Robert's patches). The
only catch is that some c++ files only compile with -O or -O0. With more
aggressive optimization the SCO "as" core dumps with SIGSEGV! The assembler
files are 10000 to 30000 lines long, so I don't post them unless you
ask for it :).

Thanks to you all,
Joao

PS: Why can't I build egcs-1.0 with --enable-shared?
I got a "-mcoff not compatible with -fpic" or something similar.
Start another thread?

-- 
Joao Cardoso, INESC  |  e-mail: jcardoso@inescn.pt
R. Jose Falcao 110   |  tel:    + 351 2 2094345
4050 Porto, Portugal |  fax:    + 351 2 2008487