Re: g77 in egcs-1.1.1

[Dave Love <d dot love at dl dot ac dot uk>]

>>>>> "JL" == Josip Loncaric <josip@icase.edu> writes:

 JL> If you mean "Red Hat Linux 5.2 (Apollo) General Errata" or "Red Hat
 JL> Linux/Intel 5.2 (Apollo) Errata", these updates affect mostly security
 JL> issues etc. and not complex arithmetic.

No, I mean stuff on rhcn.redhat.com.  I don't mean that anything has
been fixed, just that there's a built LAPACK there.

 JL> By the way, the options "-m486 -malign-double" are briefly documented in
 JL> g77 and gcc info files, 

That looks like a doc fossil bug, thanks.

 JL> compiler accepts them, but appear to be unavailable based on the
 JL> output of "g77 -v --help".  Is this consistent?

It looks as though that has since been fixed.

 >> Well, they're not Fortran77 and AFAIK relatively few f77 packages
 >> seem to fall foul of them.  AFAIR that stuff only occurs in the
 >> lapack test routines.

 JL> My view on this differs, based on what I've seen in practice.

Yes, what I meant is that in LAPACK I think it's only in the test
routines.  It is standard f90.

 JL> Sorry, no funds, only some help in a cooperative effort.  

Fine!  Thanks for offers.

 JL> P.S.  By the way, Portland Group's pgf77 has a useful option
 JL> "-fast" which specifies all typically useful optimization
 JL> options.

I don't know that compiler, but `fast' sometimes means
semantics-breaking.  The extra options that may be useful in addition
to -O2 or -O3 are probably hit-and-miss.  For instance, it's said that
loop unrolling shouldn't be the default even for Fortran for reasons I
forget.