Re: g77 - gfortran transition and AMD64 g77 bugs

[Matthias Klose]

Kevin B. McCarty writes:
 Gfortran claims not to be completely ready for use as a g77 replacement
 yet (and someone who has attempted to compile Cernlib with it reports a
 large number of problems yet).  But eventually that day will come... we
 should have some transition plan in mind by then.

upstream does explicitely ask distributors to package g77 in addition
to gfortran. etch will ship with a g77, based on g77-3.4.

 In addition, (speaking as Cernlib Debian maintainer) I have a specific
 question about g77 in Debian.  There are two g77 bugs that can cause
 significant problems in mixed FORTRAN/C applications on AMD64 machines:
 
 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15397
 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17541
 
 Could someone tell me whether it is likely that both will eventually be
 fixed in the g77-3.4 available in Debian?  If so, in what timeframe?
 The latter is fixed upstream, but the former is apparently unfixed.  It
 can apparently be worked around with the -fno-f2c flag.  But this (from
 the first half of my email) will be incompatible with any current
 FORTRAN libraries in Debian.

bugs fixed upstream will be fixed in Debian with next sync from the
CVS branch, likely to happen, when the current gcc-3.4 packages enter
testing. Other bugs are likely to remain unfixed.

Given the differences in g77 and gfortran, you may want to propose a
policy for Debian packages, how to build the packages.

  Matthias


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]

g77 - gfortran transition and AMD64 g77 bugs

[Kevin B. McCarty]

Dear list,

[CC'ed to debian-science since I imagine that's where most of the
FORTRAN users are.  Please follow up to -devel.]

With the latest g++ ABI transition soon to be finished, I have to ask
whether a similar transition is planned for g77 to gfortran.  It appears
that the ABI of code generated with the two compilers is different.
Both compilers have a -ff2c / -fno-f2c option pair controlling whether
the code is ABI-compatible to that generated by f2c.  But in g77, the
default is -ff2c, while it is the opposite for gfortran.

From the info manual for gfortran:

 _Caution:_ It is not a good idea to mix Fortran code compiled with
 `-ff2c' with code compiled with the default `-fno-f2c' calling
 conventions as, calling `COMPLEX' or default `REAL' functions
 between program parts which were compiled with different calling
 conventions will break at execution time.

 _Caution:_ This will break code which passes intrinsic functions
 of type default `REAL' or `COMPLEX' as actual arguments, as the
 library implementations use the `-fno-f2c' calling conventions.

Gfortran claims not to be completely ready for use as a g77 replacement
yet (and someone who has attempted to compile Cernlib with it reports a
large number of problems yet).  But eventually that day will come... we
should have some transition plan in mind by then.


In addition, (speaking as Cernlib Debian maintainer) I have a specific
question about g77 in Debian.  There are two g77 bugs that can cause
significant problems in mixed FORTRAN/C applications on AMD64 machines:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15397
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17541

Could someone tell me whether it is likely that both will eventually be
fixed in the g77-3.4 available in Debian?  If so, in what timeframe?
The latter is fixed upstream, but the former is apparently unfixed.  It
can apparently be worked around with the -fno-f2c flag.  But this (from
the first half of my email) will be incompatible with any current
FORTRAN libraries in Debian.

regards,

-- 
Kevin B. McCarty [EMAIL PROTECTED]   Physics Department
WWW: http://www.princeton.edu/~kmccarty/Princeton University
GPG: public key ID 4F83C751 Princeton, NJ 08544


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]