Re: [Gomp-discuss] Somethings to think about ....

[Pop Sébastian]

Hi everybody,

On Mon, Mar 10, 2003 at 04:08:53PM +0100, Steven Bosscher wrote:
> >    2. the tree modifications are not that hard, but have to be carefully 
> > planned in order to be efficient and extendable. Still they have to be done.
> 
> Agreed.  We're doing IR design here, it's the kind of stuff you should
> do "right" on the first try...  I have never seen an IR with concurrency
> awareness; Diego's probably the only one of us who has???
> 
I have seen one and it's becoming a standard!
  http://www.ecma-international.org/publications/standards/ecma-335.htm

That means that lots of people have worked on this IR.  I have asked Stallman 
about possible licensing problems, and still wait for his approval...
I'm considering to implement a CIL front-end to GCC: that will make code 
produced by Mono C# compiler run faster (since it will be native code) 
  http://www.go-mono.org/c-sharp.html
and more important will refine our GENERIC implementation.  I really think that 
for the moment what we call GENERIC is what came out from C and C++ without 
worrying much about other languages as Fortran, Java, Ada, Pascal, ...
On the other hand CIL was designed in order to support even functional languages
and was designed by a much bigger community whose interests were not only 
imperative languages...  

Well, sorry it's a bit off topic on this mailing list, but nevertheless since 
the subject on IR come up, this shows that there are standards that exist in 
intermediate representations that we will have to consider before implementing 
the GENERIC bytes on concurrency.

> Ref. (3).  I really hope Seb. comes up with a nice __builtin_pragma
> implementation *before* we start hacking the middle end.
> 
Still working on it, and yes it's weird.  These bytes are however only the top 
of 
the iceberg...


        Sebastian