[Patch, fortran] PR35680 - [4.3/4.4 regression] ICE on invalid transfer in variable declaration

[Paul Richard Thomas]

Dominique,
.
Thanks for taking a look at the patch.

> Your patch replaces the ICE by the expected error. However I have found that it
> interferes with the Daniel Kraft's patch in

That's a relief!  I had one posting recently that did not fix the PR:-)
>
> http://gcc.gnu.org/ml/fortran/2008-09/msg00411.html
>
> After some refactoring of this patch I have been able to combine the two.
> In this case I have found that the third test in the Daniel's patch gives
> only one error instead of two. I think it comes from the fact that
> sym->attr.error is not reset after use (it may also be used unitialized?).

The symbol_attributes are cleared on allocation.  I did not anticipate
resetting the flag after use - it is meant to indicate that symbol is
stuffed.  However, this is not right, is it?  It is incorrectly
referenced and it is really the expression that is not right.  I will
think about this before committing, even though this is the last
non-darwin9 regression.

On the remaining two regressions:  I see that I inadvertently fixed
PR37319.  Can we clear it, a far as you are concerned?  On PR36463 -
what do we do about it?  It would be very nice to have a clean sheet
for regressions.

Cheers

Paul