Restrict mangling.

[Coleen Phillimore]

Jim Dehnert wrote:
> 
> > Date: Mon, 19 Jun 2000 16:02:29 -0400
> > From: Coleen Phillimore <coleen at zko.dec.com>
> >
> > There's a CV-qualifier mangling for 'restrict', but restrict doesn't mean
> > anything unless it applies to the pointer itself (not what is pointed to).
> > MS and g++ from what I can tell do not overload on the basis of restrict.
> > Does it make sense to add to the mangling?
> >
> >   <CV-qualifiers> ::= [r] [V] [K]       # restrict (C99), volatile, const
> 
> Well, it seems to me that it should be part of the mangling if and only
> if it is also a factor in overloading.  Since C++ doesn't yet support
> restrict, we don't have a definitive resolution of the latter.  So, unless
> we want to specify it ourselves, and I think that would be very premature,
> the answer is up in the air.
> 
> Perhaps we should just clarify in the document that it is specified for
> use IF restrict enters into overloading decisions, and should be
> omitted otherwise, without specifying the choice.  In effect, its use
> becomes implementation-specific, as is the overloading effect of the
> non-standard restrict qualifier.
> 
> Comments?

Yes that sounds like a good resolution.  I was wondering if it should
be overloadable, actually, but that question is for the ANSI C++ committee
if they add the qualifier.  It makes sense to keep it in the ABI mangling.

Thanks,
Coleen
> 
> Jim
> 
> -           Jim Dehnert         dehnert at sgi.com
>                                 (650)933-4272

-- 
-----------------------------------------------------------------------
Coleen Phillimore                  | mailto:coleen at zko.dec.com
Compaq Computer Corp.   Nashua, NH | COMPAQ C++ Compiler Development
-----------------------------------------------------------------------