Peter,
On 2005 Mar 31 , at 15.16, Peter W. Draper wrote:
> I've also set my Alpha build away, but that hasn't been getting very
> far.
The Alpha build failed at RAL last night, too, because the test for
'f77 -E' passed spuriously (I've no idea what -E does here, since it
seems not to do `just preprocessing', and it isn't documented). I'm
fixing that.
> The main problem seems to be with the compiler version I have. When I
> run
> .F files through it says:
>
> cpp: Severe: No such file or directory
> ... file is '`-DLANGUAGE_FORTRAN_95'
> f90: Severe: The input stream is empty
>
> (which isn't seen at RAL). Adding the -old_f77 flag to get a proper F77
> compiler works.
Right, I'll see if I can sanely add '$FC -old_f77' to the list of
invocations which are attempted. Failing that, it can go in the
per-platform build notes.
> However that then produces an error with the
> AC_FC_RECL_UNIT test, which looking in config.log says:
>
> fort: Error: conftest.f, line 24: Unknown keyword value
> : STATUS = 'REPLACE',
>
> so that's not surprising. There is no 'REPLACE' status in F77 (just
> 'OLD',
> 'NEW', 'UNKNOWN' and 'SCRATCH'), so you'll need to re-work that test.
Ta -- I missed that.
> Meanwhile I'll lobby for a compiler upgrade (not much chance).
Or you could lobby to avoid the compiler being upgraded: (a) it sounds
like this would be successful (and thus productive of a warm glow), and
(b) antique compilers are to some extent useful in finding problems
like the REPLACE one above.
See you,
Norman
--
----------------------------------------------------------------------
Norman Gray : Physics & Astronomy, Glasgow University, UK
http://www.astro.gla.ac.uk/users/norman/ : www.starlink.ac.uk
|