error/warning message

[Brooks Moses]

Anglade Pierre-Matthieu wrote:
> I may mistake but in the gfortran site I can't find anywhere a mailing
> list or equivalent  tool, dedicated to bug reporting or so. I
> acknowledge that gfortran developers may not have time for improving
> features like output messages and that it would be better to improve
> them myself and send patch...  But I don't feel like beeing able to do
> that at the moment.

Since gfortran is part of GCC, bugs should be reported using the main 
GCC bug-reporting database.

This should probably be explained more clearly on the gfortran home 
page, so thank you for mentioning this.  (I assume that by "the gfortran 
site" you mean that you looked around the page at 
http://gcc.gnu.org/fortran/ and the pages linked from it, but did not 
look at http://gcc.gnu.org/?)

> I'd be very glade if someone can tell me about the right way to report
> such problems to gfortran developpers.
> 
> Just for instance, my last problem with an error message:
> gfortran:
> Error: MODULE attribute conflicts with PROCEDURE attribute at (1)
> ifort:
> fortcom: Error: eneofrho_tfw.F90, line 64: The name of the module
> procedure conflicts with a name in the encompassing scoping unit.
> 
> Thought gfortran message is perfectly relevant I was not able to find
> the error...

Francois-Xavier mentioned the web page for the GCC bug-reporting system. 
  A couple of things that he didn't mention, which would be useful to 
include when you do make bug reports:

* If possible, include a small sample of code that can be compiled to 
produce the problem you're talking about.

* Include the complete text of the error messages (including the part 
where it quotes the line of code that contains the error).

* Describe, as clearly as possible, what you think would improve the 
error messages.  Including the "better" example from ifort is useful, 
but it would also be helpful to have specific suggestions for what you 
think could be done.  For instance, in this case, is the problem the 
fact that gfortran is explaining this as two incompabile attributes of 
the same name where ifort is explaining this as two names that conflict? 
  Or is it just that the wording is short and doesn't contain enough 
detail -- and, if so, what particular detail would help?

Thanks!
- Brooks