Old app recompiled with f77 failed but succeed with g77

Question

hello all,
i need to recompile an old app written in fortran77, but it fails at the execution:
Starting nonmem execution ...
Note: IEEE floating-point exception flags raised:
    Inexact;  Underflow;  Invalid Operation;
See the Numerical Computation Guide, ieee_flags(3M) after generating a core, dbx tells me:
t@1 (l@1) program terminated by signal FPE (invalid floating point operation)
Current function is cels
  103         IF (ICAT.NE.0.AND.CTLW.GE.CTUP) THEN
(dbx) where
current thread: t@1
=>[1] cels(cnt = 0.0, p1 = ARRAY, p2 = ARRAY, ier1 = 0, ier2 = 0), line 103 in "CELS.f"
  [2] ncontr(cnt = 0.0, ier1 = 0, ier2 = 0, l2r = 0), line 329 in "NCONTR.f"
  [3] obj(o = -NaN.0), line 1636 in "OBJ.f"
  [4] zxmin1(funct = 0x80e8c10, nsig = 3, maxfn = 450, jopt = 0, w = ARRAY, opint = 5), line 222 in "ZXMIN1.f"
  [5] estim(ier = 0), line 156 in "ESTIM.f"
  [6] MAIN(), line 45 in "NONMEM.f"and:
(dbx) print ICAT
icat = 3
(dbx) print CTLW
ctlw = NaN
(dbx) print CTUP
ctup = -NaN.0what i don't understand is that the same sources compiled with g77 (2.95) works without problems. That is g77 is able to set ctlw and ctup correctly!?
Is there a flag in f77 to mimic the g77 behaviour?
here are the flags i'm using:
f77 -g -C -fnonstdthanks in advance for help,
gerard
Edited by: mister_g on 24 juin 2009 09:27

Answer 1

it seems that the problem is related to COMMON variables, as they are strangely declared, but as i said before, it's a "commercial" app, even if this is the responsibility of the customer to recompile it.
What we observed is that, before entering into a function, a variable decalred in a COMMON has a correct value, but in the function, it contains NaN. In the earlier releases, according to the official docs, developers used to use sun compiler, so my guess is that something has changed in Sun f77 since the old times?
i gave a try with old workshop WS6.u2, the program failed too, but i'm unable to generate a core or to tell the debugger to stop where the error occured:
(/opt/SUNWspro.old/bin/../WS6U2/bin/sparcv9/dbx) catch
HUP INT QUIT ILL TRAP ABRT EMT FPE BUS SEGV SYS PIPE ALRM TERM USR1 USR2 CLD PWR WINCH URG POLL STOP TSTP CONT TTIN TTOU VTALRM PROF XCPU XFSZ WAITING LWP FREEZE THAW CANCEL LOST XRES 39 40 RTMIN RTMIN+1 RTMIN+2 RTMIN+3 RTMIN+4 RTMIN+5 RTMIN+6 RTMIN+7
(/opt/SUNWspro.old/bin/../WS6U2/bin/sparcv9/dbx) run < FCON > REPORT5
Running: nonmem < FCON > REPORT5
(process id 25314)
Reading libc_psr.so.1
Note: IEEE floating-point exception traps enabled:
    overflow;  division by zero;  invalid operation;
Nonstandard floating-point mode enabled
See the Numerical Computation Guide, ieee_handler(3M), ieee_sun(3M)
execution completed, exit code is 0
(/opt/SUNWspro.old/bin/../WS6U2/bin/sparcv9/dbx) where
dbx: program is not active
(/opt/SUNWspro.old/bin/../WS6U2/bin/sparcv9/dbx)Thanks in advance for help,
gerard