iraf.net

The answer to this question may be that its not supported but I'm
puzzled as to what I found recently when I tried to build IRAF under
Centos 4.1 This is what I found. IRAF built and installed with no problems at
all. I then started to build some packages I needed. Fitsutil and color
were again no problems. When I came to gmisc (and I had similar problems
with nmisc) I got messages in the spool file that dcgms_ and some other
routines were being called with inconsistent numbers of parameters when
compiling t_dispcor.x Sure enough, if I looked at the source of
t_dispcor.x there were places where the routine was being called with 6
parameters and places where it was being called with 8. From the
declaration I would have expected it to have 8. Previously I had built these packages under Red Hat 9.0 with no
problems so I ran xc with the -F switch and compared the two .f files
generated on the two OS's. The fortran generated was identical on the
two. I then did a bit of digging around and worked out that the f2c
compiler seemed to be using the switches -ARw8 -Nn802. Since I had the
t_dispcor.c file on Centos left over after the build had bombed out, I
ran f2c on RH9.0 with the above switches and compared the two C code
files. Once again they were identical and indeed dcgms_ was called in
different places with different numbers of parameters. The mystery to me is how the system built apparently correctly under
RH9.0 rather than why it fails under Centos. It appears under Centos
that the compiler is correctly pointing out that the routine is being
called inconsistently. Any pointers as to what is going on would be gratefully received. Many thanks Chris--
_______________________________________________________________________
Observatory Sciences Ltd. Email: cjm@observatorysciences.co.uk
William James House Tel: 44 - (0)1223 - 508257
Cowley Road, Fax: 44 - (0)1223 - 508258
Cambridge, CB4 0WX, UK http://www.observatorysciences.co.uk

Hi Chris,
It appears that some of the newer GCC versions are a bit
stricter about prototyping and creating errors where they used to
be ignored. I couldn't find a flag to disable this checking but
you can work around it by using G77 as the compiler rather than
the F2C/f77.sh script. To do this, define setenv XC_F77 g77Put this in your .cshrc file (or the equivalent if you use bash) so
it will be picked up by the mkpkg/xc properly and it should allow
the mismatched arguments.
We'll patch the code for future releases and this is only
a temporary workaround. For packages which still won't compile then
using prebuilt binaries should also work (let us know if you can't
find binaries for a particular package). Hope this helps.Cheers,
Mike Fitzpatrick