PR debug/19345

[Daniel Berlin]

On Fri, 11 Mar 2005, Jason Merrill wrote:

> On Tue, 08 Mar 2005 10:28:44 -0500, Daniel Berlin <dberlin@dberlin.org> wrote:
>
>> However, according to Jakub,
>> "
>> TYPE_NAME (TYPE_MAIN_VARIANT (origin)) on that testcase is NULL, so it
>> doesn't help match."
>
> ...which is why we still need TYPE_STUB_DECL.  It seems to be the only way
> we can reliably find a decl for a RECORD_TYPE.  Except that in this
> testcase we can't.
>
>> Since this was our proposed solution, i think the only option we have at
>> this point is to remove the type = TYPE_STUB_DECL (type) line, and just
>> ignore TYPE_STUB_DECL.
>
> No, that would completely break that piece of code, which is trying to find
> the right function to force out.

Sigh.
>
>> Either that, or someone needs to fix the "real underlying bug" here, if
>> we can figure out what that is.
>
> I think the real bug is that TYPE_STUB_DECL isn't being set for the
> RECORD_TYPE.

Hmmm.
I don't know when i'll have time to try to figure out why this is the 
case.


>
> BTW, why isn't this cc'd to a mailing list?

Accident.  I thought i did.
It is now.

>
> Jason
>