Re: [PATCH] New pragma handling mechanism.

[Mark Mitchell <mark at codesourcery dot com>]

Richard Henderson wrote:
> On Tue, Apr 12, 2005 at 04:40:09PM -0700, Mark Mitchell wrote:
>
> > I'd really like to avoid the pragma handlers seeing subsequent file 
> > content.
>
>
> Ok, then what do you suggest for handling the rather tight coupling
> of open mp pragmas with the immediately succeeding statement?

I tried to explain what I had in mind here:

http://gcc.gnu.org/ml/gcc-patches/2005-04/msg00934.html

but, obviously, it was either unclear or not compelling.

If you give me a hint as to which, I'll try to rectify the problem. :-) 
 (Though, as I'm off email again tomorrow, I might be a bit slow to do 
so.)

The key idea is that the parser would, prior to parsing a statement, or 
perhaps after, if more convenient, ask "are there any #pragmas that 
apply to this statement?", and then take appropriate action.

Certainly, if you do this before the statement, there is nothing that 
the parser cannot do that the #pragma handler could do.  However, the 
additional structure might allow the parser to do things the handler 
cannot easily do -- like, in a C++ template, squirrel away the #pragma 
until instantiation time.  I know it's possible to implement OpenMP 
using such a system, in that I have seen a worked example.

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