Re: PROPOSED PATCH: Robustify cp_fold_obj_type_ref() for ObjC++ use

[Mark Mitchell <mark at codesourcery dot com>]

Andrew Pinski wrote:

> On Sep 21, 2004, at 12:00 AM, Mark Mitchell wrote:
>
> > > Why make an extra hook?
> >
> >
> > Because that will keep more of the Objective-C++ code in the 
> > Objective-C/C++ subdirectory, where it ideally belongs.  Thinking of 
> > hooks like virtual functions, what you want to do is derive from the 
> > C++-compiler base class.
>
>
> If that is so, then really we should not be checking the language
> dialect at all in the C front-end for C++ but we don't so because
> of speed reasons, we do we have to it here.

Are you saying that introducing a hook here will result in a measurable 
slowdown in the Objective-C++ front end?  Do you have evidence of that?

At a meta-level, I don't understand what you are trying to accomplish 
with this discussion.  I reviewed the patch, and suggested a way of 
solving the problem that I would prefer to the way in which Zem did it.  
If my suggestion is technically infeasible, then we should look for 
another way.  If you think a hook is a horrible idea, then you can ask 
my fellow C++ maintainers; perhaps they will approve the original 
patch.  But, unless you really feel that a hook would be a horrible 
implementation approach, it seems like it would be simplest for you 
and/or Zem to implement the hook and move on with things.

--
Mark Mitchell
CodeSourcery, LLC
(916) 791-8304
mark@codesourcery.com