Re: [fortran, patch] Fix conflict detesction for VOLATILE attribute (PR30520)

[Brooks Moses <brooks dot moses at codesourcery dot com>]

Jerry DeLisle wrote:
> See wording here:
>
> Tobias Burnus wrote:
>
> > > !   /* TODO: The following allows multiple VOLATILE statements for
> > > !      use-associated variables and it prevents setting VOLATILE for a host-
> > > !      associated variable which is already marked as VOLATILE in the host.  */
>
> I did not go back and look at the rest of the thread to see if you fixed this 
> already.

So far as I know, no.  Tobias and I talked about this on IRC when he was 
creating the patch, and it's going to be particularly tricky to solve. 
The problem is that, currently, gfortran assumes that if a given 
attribute is set on a variable then setting it again is an error. 
However, this is sometimes true and sometimes not true with VOLATILE, 
and thus fixing this TODO will require adding machinery for tracking 
where the attribute was set.

The current patch was a compromise that uses some reasonable guesses to 
allow the most common sorts of multiple declarations of VOLATILE, while 
rejecting what I would guess to be the most common forms of incorrect 
code.  The TODO comment documents where those guesses fall short.  I 
think it's a reasonable patch to put into place while waiting for a 
complete solution.

- Brooks