Re: [Gomp-discuss] Release candidate

[Lars Segerlund]

 Good work !

  I think we need to start to have a look at what have to go in the tree-ssa 
form ( if we use something like in diegos dissertation ).

 The reason for this is that we have to get the stuff through the frontends, 
and it will take some time to figure out how , ( I suspect ).

 So basicly we have to start to think about what can be bound to a block of 
code and how interrelationships.

 I'll do a dig and have a really good look at diego's stuff.

 / Lars Segerlund.


On Mon, 1 Nov 2004 11:38:31 +0200
Biagio Lucini <address@hidden>(by way of Biagio Lucini  <address@hidden>) wrote:

> Hi All - sorry for not being able to provide updates during the last week. I
> was dealing with the autostuff and caught in autohell, as usual. Now
> I have a tarball that should compile when dropped into $srcdir, but before
> committing the patch I need a cross-check. So below is the procedure I have
> followed (wich, due to my lack of experience with gcc-cvs is likely not to be
> the most appropriate):
> 
> a) check out a fresh version of the gomp branch
> cvs -d :pserver:address@hidden:/cvs/gcc -z 9 update -P -rgomp-branch gcc
> b) cd gcc
> c) tar xvzf libgomp.tar.gz
> d) cd ..
> e) mv gcc src
> f) src/configure --enable-languages=c
> g) make bootstrap
> 
> The bootstrap succeeded on a i686 Debian 3.0.
> 
> If the procedure works without errors for someone else, I could probably
> submit libgomp as a patch. Comments about the conformity of coding style with
> gcc standard are also very much appreciated. For a list of things in which
> someone could help please see the TODO file (which is largely incomplete and
> just deal with this current rudimentary stage). To to pressing family
> committents (my second son is expected in mid November) I will disappear for
> an unquantifiable period of time, so it would be nice if someone could do
> some little work on the sources in the meanwhile. The OpenMP standard is
> pretty much self-explicative, so there should not be much doubt on what it is
> needed in the runtime part of the library. We could start with the POSIX
> implementation of it, for instance.
> 
> Another blocking issue can be the copyright: Diego and Jacob, any news from
> that front?
> 
> I hope to hear constructive and (why not) destructive comments very soon
> (even better if all come with patches and solutions :-))
> 
> Cheers
> Biagio
> 
> P.S.: The tarball is called libgomp.tar.gz and can be downloaded from
> http://www.phys.ethz.ch/~lucini/GOMP
> 
> --
> =========================================================
> 
> Biagio Lucini
> Institut Fuer Theoretische Physik
> ETH Hoenggerberg
> CH-8093 Zuerich - Switzerland
> Tel. +41 (0)1 6332562
> 
> =========================================================
> 
> 
> _______________________________________________
> Gomp-discuss mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/gomp-discuss