Hi,
> On 05/26/2011 02:30 PM, Paolo Carlini wrote:
>> We want to do that only when processing a template, because otherwise we
>> get a duplicate diagnostic, see, eg, auto9.C
> Hmm, where's the error coming from in the non-template case?  From 
> cp_build_c_cast?  In that case always giving the error in 
> build_functional_cast and then returning error_mark_node should avoid 
> duplication.
I should have told you. The error is produced via 
complete_type_or_maybe_complain called from line 1650 of 
build_functional_cast itself, by cxx_incomplete_type_diagnostic.
>> error_mark_node unconditionally, means a better diagnostic, without
>> redundant "array bound is not an integer constant before...".
> I'd rather deal with that by suppressing that error if we already have 
> error_mark_node.
complete_type_or_maybe_complain, when type is an error_mark_node 
understands that an error has been produced already and simply returns 
NULL_TREE. Thus, something like the below works. Is it closer to what 
you had in mind?

Paolo.

/////////////////