Re: Is passing via rank-1 assumed-size array legal?

On Apr 27, 4:55 pm, steve <kar...@xxxxxxxxxxx> wrote:
Given you have access to non-PGI compilers, it would
be prudent to use their debugging features.  With
gfortran this would be use of the -Wall -fcheck=all

I have these (and similar for remaining compilers) turned on
during development. The thing is, the problem does not occur
with other compilers, so the availability of other compilers
doesn't help me.

  Finally, I'm not saying the code was fine with F77 and is
acting up only now -- the original code was never compiled
with this particular compiler.

Well, this is a porting issue.  One should always obtain
baseline runs before altering code.  Preferably, testing
as many (if not all) execution paths through the code.

You are right. I did what you call "baseline runs", but not
with all compilers -- just Sunpro and gfortran.

- J.

Relevant Pages

  • Re: Compilation problem with gfortran
    ... I'm trying to compile some code of mine with gfortran. ... Argument 'dydx' of pure subroutine 'rkckcontrolstep' at must ... Sun compilers accept this, but I appreciate that that isn't a definitive ...
  • Re: gfortran/ifort format issues in files
    ... while the gfortran version uses LF only. ... If both compilers can read files with either terminator, ... subroutine read_line (infile, line, ios)! ...
  • Re: OpenMP "not working" on gfortran
    ... the OpenMP code I wrote works on Mac OS X gfortran 4.5.2. ... and both of these compilers result in no speedup. ...
  • Re: Suspicious Gfortran
    ... Fortran 77 compilers that exist, and thus doesn't mean anything in this ... problem is coming from a bug in your code, rather than a bug in gfortran. ...
  • Re: use std=f2003 or std=f95?
    ... what do YOU mean by 'legacy'? ... Of which gfortran supports very few. ... asymmetric range is just asking for trouble. ... A LOT of compilers will negate values when it ...