Re: [Gomp-discuss] Re: [gomp] parse OpenMP directives

[Lars Segerlund]

 On a different note, before we have working parsers in place we can do 
nothing, so I think getting working parsers are a good start.
 When we can parse ( and discard the info as far as gcc is conserned ), we can 
start bringing the info throught the frontends and into tree's ( se diegos 
paper ).

 Most frontends have a scanner and a parser part, so our first is to get the 
omp directives recognized in the scanner, when this is done we have to bring it 
to the parser and so on .. logical progression.

 My point is, let's get scanners/parsers for FORTRAN/C ( perhaps C++ later ) 
and were off to a good start.

 i.e. postpone the 'BIG' problems until we must start handling them and code 
now.
 
 / Lars Segerlund.

On Wed, 13 Oct 2004 11:52:37 -0400
Scott Robert Ladd <address@hidden> wrote:

> Sebastian Pop wrote:
> > The following patch registers a set of functions for parsing the
> > OpenMP directives.  Most of the functions are not yet implemented, but
> > I wanted to get some feed-back from the CPP people on whether this is
> > the right approach for parsing the directives.
> 
> The directives can not be simply translated into function calls. OpenMP
> effectively "rewrites" the program in order to support variable scoping,
> locking, and thread creation; in essence, the parser needs to analyze
> the code within the OpenMP block and write new, equivalent, parallel
> code. I posted examples of this earlier.
> 
> ..Scott
> 
> -- 
> Scott Robert Ladd
> site: http://www.coyotegulch.com
> blog: http://chaoticcoyote.blogspot.com
> 
> 
> 
> _______________________________________________
> Gomp-discuss mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/gomp-discuss