All (but particularly Al and Brad),
On 2004 Jul 31 , at 21.03, Norman Gray (GLA) wrote:
> Add private macro _STAR_RESTFP_FIXUP, which addresses the
> restFP/saveFP problem on OSX (for good?). This is automatically
> invoked from STAR_DEFAULTS.
> This renders macro/hack STAR_FC_LIBRARY_LDFLAGS obsolete -- redefine
> this so that it causes an autoconf error if it is used.
After some discussion on the [log in to unmask] list (the
thread beginning
<http://lists.apple.com/mhonarc/fortran-dev/msg00768.html>, if you're
interested, Al and Brad), I think I have sorted the restFP/saveFP
problem on OSX. I've therefore retired the STAR_FC_LIBRARY_LDFLAGS
hack.
It should now be the case that links between object code compiled with
(Apple) gcc and (OpenDarwin or Fink) g77 should now work, not only in
the case where you're knowingly linking one with the other, but also in
the case (the one which came up when I was trying to build ASTROM) when
you're linking gcc-compiled code using gcc, but linking against
libraries built using g77.
This will produce some extra chatter in the ./configure run when
building on OSX, but should produce nothing noticeable on other
platforms. Let me know if this is not the case.
*ALL*: I have therefore removed the STAR_FC_LIBRARY_LDFLAGS invocation
in ast's configure.ac, and changed it to standard AC_FC_LIBRARY_LDFLAGS
in idi, nbs, convert/idl, icl, and dipso (I couldn't link and test this
last one, because I haven't got the right dependencies built right now,
but I'm sure it'll be OK). ccdpack still has STAR_FC_LIBRARY_LDFLAGS,
but I know I can't currently build that, so I didn't try. Mark: can
you make this change.
Apart from dtask, which I have other things to fix in (thanks for those
pointers, Tim), I _think_ these are the only components which used the
hack.
If you have STAR_FC_LIBRARY_LDFLAGS in anything else, then I've
modified starconf so that autoconf will fail, and so force you to
either change or delete this.
> Do the same with long-obsolete STAR_HAVE_FC_OPEN_READONLY, which used
> AU_DEFUN, and so merely whined quietly rather than objecting.
I presume noone still has this macro.
See you,
Norman
[I think we're getting there...]
--
----------------------------------------------------------------------
Norman Gray : Physics & Astronomy, Glasgow University, UK
http://www.astro.gla.ac.uk/users/norman/ : www.starlink.ac.uk
|