Sorry about the very belated reply. Finally managed to solve the
problem. Thanks to Brian Lavelle, James Sainsbury, Dr Thomas Blinn, Corey
Brenner, Paul Roetman and Jim Barriault.
As I suspected, the _FSTAT_64 string is used in version 5 of the OS. The
problem was, it was cropping up on our version 4 machines. The user swore
the code use to compile under version 4 and that they had not done anything
to it. This was partially true.
After much wailing, gnashing of teeth and use of the nm command (Thanks
Paul - I didn't know that existed) I finally got the problem down to a
single library file. When I asked the user where he got the file from, he
was quite happy to admit that he didn't compile it - he had used the same
version that one of the other people use. And yes, thats right, that
version was compilied up on version 5.
Recompiling the library file for version 4 fixed the
problem. Argh!!!! So easy.
Thanks for all your help,
Ryan.
(Original query follows)
I'm not sure if this is an OS issue or a programming issue, but this list
is about the closest thing I can find to the correct forum. One of our
users recently tried to compile one of his programs. Below is a copy of
the output.
ord:/home/pg/meulener/roms/roms_src>make
cpp -P -DSERIAL -DDECALPHA -I/home/pg/meulener/roms/roms_src wrt_avg.F >
wrt_avg.f
cpp_clean wrt_avg.f
f90 -c -fast wrt_avg.f
rm wrt_avg.f
f90 -fast -o roms roms.o initial.o main2d.o main3d.o
analytical.o bulk_flux.o bvf_mix.o checkdefs.o close_io.o
def_avg.o def_his.o def_info.o def_rst.o
def_station.o def_var.o diag.o exchange_2d.o exchange_3d.o
fill_2d.o fill_3d.o get_2dfld.o get_3dfld.o get_cycle.o
get_data.o get_date.o get_grid.o get_initial.o
get_mask.o get_tile.o horz_mix.o init_arrays.o
init_scalars.o inp_par.o lenstr.o lmd_bkpp.o lmd_skpp.o
lmd_swfrac.o lmd_vmix.o lmd_wscale.o metrics.o
mp_routines.o
my25_corstep.o my25_prestep.o nf_fread.o nf_fwrite.o
obc_volcons.o oi_update.o omega.o opencdf.o output.o
pre_step3d.o prsgrd.o q3dbc_im.o rho_eos.o rhs3d.o
set_2dfld.o set_3dfld.o set_avg.o set_data.o
set_depth.o set_massflux.o
set_nudgcof.o set_scoord.o set_vbc.o set_weights.o sg_bbl.o
step2d.ostep3d_t.o step3d_uv.o t3dbc_im.o t3dmix_geo.o
t3dmix_iso.o t3dmix_s.o timers.o u2dbc_im.o
u3dbc_im.o u3dmix_geo.o u3dmix_s.o v2dbc_im.o v3dbc_im.o
v3dmix_geo.o v3dmix_s.o w3dbc.o wrt_avg.o
wrt_his.o wrt_info.o wrt_rst.o wrt_station.o wvelocity.o
zetabc.o -L/home/pg/meulener/roms/roms_src -lnetcdf
ld:
Unresolved:
_F64_fstat
f90: Severe: Failed while trying to link.
*** Exit 1
Stop.
Obviously it failed, because of what looks like a missing library and so
it got heaved over to me to try and figure out what the problem is. I
have since discovered that the version of the fortran compilier does not
appear to make any difference, but the version of the OS does.
On all of the Tru64 version 4.0x machines I tried it on it failed, but
trying it on 5.0, 5.0A and 5.1 machines worked fine.
Does anyone know if its possible to fix this problem without having to
upgrade all of our machines to version 5?
Thanks,
Ryan.
--
Ryan McConigley - Systems Administrator _.-,
Computer Science University of Western Australia .--' '-._
Tel: (+61 8) 9380 7082 - Fax: (+61 8) 9380 1089 _/`- _ '.
Email: Ryan_at_cs.uwa.edu.au - http://www.cs.uwa.edu.au/~ryan '----'._`.----. \
` \;
"You're just jealous because the voices are talking to me" ;_\
Received on Wed Apr 04 2001 - 08:19:21 NZST