roebuck at mdanderson.org
Mon Jan 30 17:15:45 CET 2006
On Mon, 30 Jan 2006, Jan de Leeuw wrote:
> On Jan 30, 2006, at 07:04 , Paul Roebuck wrote:
> > On Sat, 28 Jan 2006, Simon Urbanek wrote:
> >> On Jan 27, 2006, at 4:54 PM, Paul Roebuck wrote:
> >>> I continue to hear Apple's branch has broken gfortran, though never
> >>> hear specifically what is broken about it.
> >> It doesn't compile - that's what's broken :). My impression is that
> >> there is quite a bunch of new pieces and changes in the C back-end
> >> code that were shared by the gfortran, but the fortran part was not
> >> updated correspondingly. I tried to patch some of the holes, but
> >> didn't get very far. Maybe if someone has the spare time to trace all
> >> the changes individually in both the FSF and Apple branch could put
> >> it together...
> >>> I don't mind filing a bug report with them too
> >> The problem is Apple doesn't support gfortran and never did. The
> >> ability to compile gfortran from the Apple branch was just incidental
> >> and required some tweaking as well (basically manually enabling f95
> >> in the build script and patching together fat gfortran libraries),
> >> but it was harmless. When you ask them, they say "use f2c" (but R
> >> doesn't support that).
> > I guess what I find odd about [gfortran not compiling] is
> > that no one seems to mention that in the Apple mailing
> > list over the last couple months.
> > <http://lists.apple.com/archives/fortran-dev/2006/Jan/>
> gfortran compiles find on powerpc systems -- not on mactel
> systems. until a few weeks ago, people with mactel systems
> were under nondisclosure and could not post the info on
> any mailing lists.
Obtaining a newer build of 'gfortran' than 5202 is not out
of the question then. Just wanted the matching F95 compiler
for Xcode 2.2.1 build (PPC).
SIGSIG -- signature too long (core dumped)
More information about the R-SIG-Mac