Common V3 coding bug

[Mark Mitchell]

--On Wednesday, July 18, 2001 12:28:32 PM +0100 Jason Merrill 
<jason_merrill@redhat.com> wrote:

> We really need to emit a warning for this case, like the existing warning
> about typename.  My strong preference would be to accept the code with an
> unconditional pedwarn; there's a LOT of code that depends on this
> behavior.

Yeah, I probably agree -- but I'm trying to just implement the standard
first; that's hard enough.  My only concern is that lookups aren't free,
but I don't yet have any data to know how un-free, and the new parser
isn't smart about this anyway, so there's probably a lot more win from
other lookup-reducing ideas.

Let's reserve these battles until I can pretend to have something that
works passably.  We might as well fix up V3 at the same time; there's
no reason for it not to be conformant.

Note that the "supporting old code" thing is the reason that the V3 
developers, conscientious folks, are writing wrong code -- the compiler 
hasn't been helpful in spotting the bugs.  So, we want a mode that is 
strict, and that is the mode I'm writing; then we can add, under a flag
(or the absence of -pedantic) whatever other gunk we need.

-- 
Mark Mitchell                mark@codesourcery.com
CodeSourcery, LLC            http://www.codesourcery.com