Re: [C++ PATCH]: remove some memory, speed up some parsing

[Mark Mitchell <mark at codesourcery dot com>]

Nathan Sidwell wrote:
> Mark,
> this patch stops us creating binfos for all but actual aggregates, and
> prevents access checking earlier in templates (we end up collecting all
> the accesses and then ignoring them later at the moment).

In the longer run, of course, we need to collect them and then do them 
at instantiation time.  Still, no reason to collect them now, only to 
throw them away.  So, this part of the patch is OK.

After your patch, how does the situation where a template type parameter 
is used as a base class work?  Do we create a binfo at that time, or do 
we not binfos in that case any more?  Oh, I see, we already create a 
binfo in xref_basetypes in that case.  OK, so, this change is OK too.

--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com