>> Still, I can't completely stifle my curiosity about the g77 problems. Is
>> there a utility out there that checks FORTRAN code for f77 compliance?
>
> The best test is to run it through f2c. f2c *is* nothing else than the
> f77 'authority', believe me.
I _like_ f2c but I would'nt go quite that far. Another tool with better
diagnostic output is
<http://www.dsm.fordham.edu/~ftnchek/>
Source code (C) and compiled executables for a number of systems are
available.
When I run ftnchek on pmx217.for with no options I get a lot of (mostly)
harmless warning messages, but
ftnchek -pretty=none -truncation=none -f77=none -f77=mixed-common *.for
gives a few error messages like
Warning in module PMXA file pmx217.for:
Common block COMMAC line 478 has mixed
character and non-character variables (nonstandard)
The offending source line is
common /commac/ macnum,mrecord,mplay,macuse,icchold,lnholdq,endmac
with character*128 lnholdq and logical endmac. Since different compilers
store character variables differently, trying to access non-character data
located in common after the character data can lead to memory errors. I
suspect that this is what you are seeing. The solution (best case) would be
to segregate character variables into their own commons, but (next best)
moving the character variables in commons to the end would probably solve
the problem.
HTH,
Eric
+------------------------------------------------------------------------+
| Eric Petersen, University of Minnesota (Home) 612-533-2654 |
| Dept of Aerospace Engineering and Mechanics 612-624-3516 |
| Minnesota Supercomputer Institute 612-624-1524 |
| [EMAIL PROTECTED], [EMAIL PROTECTED] |
| http://www.aem.umn.edu/people/students/epeterse/ |
+------------------------------------------------------------------------+
