Re: [Gomp-discuss] Open64's implementation of OpenMP

[Pop Sébastian]

On Tue, Feb 04, 2003 at 12:43:12PM -0500, Diego Novillo wrote:
> On Tue, 04 Feb 2003, Pop Sébastian wrote:
> 
> > Diego, could we keep patches for OMP in tree-ssa?
> > 
> Yes, but I'd like to avoid doing it directly on the branch.  I've
> got an alternate idea.  Let's create a sub-branch of tree-ssa for
> doing all our OMP hackery.  Two reasons:
> 
> - We will be doing changes to files in the front end.  I'd rather
>   not pollute the branch itself because we will be looking at
>   merging the branch in a few months.
> 
That's good news :-)

> - Changes for OMP can go through a looser review process.  If we
>   break bootstraps on the OMP sub-branch, we won't care too much
>   at first.  OTOH, tree-ssa is under stricter rules.
> 
> I volunteer to keep both branches in sync.  I can create a
> sub-branch for OMP whenever you guys want to start hacking the
> front end.
> 
> I also volunteer to do all the middle-end stuff.  I don't think I
> want to deal with all the front end bits.
> 
> Thoughts?
> 

Firstable, thanks for your help Diego.

Having a sub-branch for the beginning will be of great help.  
I think we'll have to merge it quickly to tree-ssa once we begin to touch 
the middle-end bits.  That could go very quickly if we decide to use a 
representation as in Open64: a single tree node (for example OMP_PRAGMA)
with some information attached to it (pragma type, and pragma arguments).
for all pragmas that are not yet lowered.  Translation of these
OMP_PRAGMA to RTL will emit a warning (something like: 
"Pragma %s not handled.".  

Once we have all this I think we could merge with tree-ssa, allowing 
analyzers to emit nodes with parallel semantics based either on the
OMP_PRAGMA nodes, or on their results.