Re: C++ PATCH: New parser

[Gerald Pfeifer <pfeifer at dbai dot tuwien dot ac dot at>]

On Mon, 30 Dec 2002, Mark Mitchell wrote:
> I'm going to be fixing the other bugs and looking at the performance
> complaints first.

Of course! Though I have to admit that, giving the invasiveness of
this change, it seems to be quite stable.

Unfortunately, I just triggered a bug; please find preprocessed source
attached to this message:

  % gccvs dl.ii
  :
  depgraph.h:661: internal compiler error: in pop_binding, at cp/decl.c:1182


> Yes.  I expect a FAQ over the next release cycle or so will be "why
> do I get weird errors in templates?"

Absolutely. I have spent some time today debugging code of colleagues
which worked fine from EGCS 1.1 to the GCC 3.3 branch and started issuing
weird errors with the new parser -- though in all cases indeed the code
was faulty, and not the new parser. ;-)

> We deprecated implicit typename a while back.

The problem is, those cases which we used to diagnose (and issue a
warning) probably have been fixed in active source bases, but the others
often have not; at least that's my guess after todays debugging sessions.

Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

Attachment: dl.ii.bz2
Description: Binary data