PR33609: ICE on arithmetic overflow

[Jerry DeLisle]

Dominique Dhumieres wrote:
>> I'm not completely sure it is a fix.  It is more like a bandaid.
> 
> I don't understand why. Could you be more explicit?
> 
> TIA
> 
> Dominique
> 
Simplification is returning a NULL expression.  The question is should we throw 
an error elsewhere or perhaps return gfc_bad_expr instead.  Its safe to bail on 
the NULL as the patch does.

I think I will commit it tomorrow, but leave the PR open with note to remind to 
look into the other possibility.

Jerry