Re: [Gomp-discuss] Plan ... coments wanted !

[Lars Segerlund]

 Ok, so one of our goals is to identify a subset that most threading 
libraries can handle ?

 This way we can use any library independent of the model behind it, 
and thus run on anything :-)

 I think this is the way forward, but I would like to postpone the 
debate about this until we have more 'meat' on the bones :-)

 This however brings up the question of what our first taget ? ( 
milestone ) should be, any ideas ?

 / Lars

Biagio Lucini wrote:
> On 29 Jan 2003, Steven Bosscher wrote:
>
>
> > Op wo 29-01-2003, om 16:04 schreef Biagio Lucini:
> >
> > I just mean to say that it may be a better plan not to rely on NPTL (as
> > good as it may be, it sure looks good) because most people just don't
> > have it yet and probably won't have it for a while either.  Therefore,
> > to say that all GNU software will go there "in the near future" just
> > seems a little too optimistic to me.
> >
> > It is an interesting discussion point which library we should target
> > first.  IMO it should be one that most people in our "target audience"
> > have at their disposal.  I guess that would be LinuxThreads, and since
> > NPTL is mostly backward compatible with that library, I would like to
> > see the first low-level runtime should be based on LinuxThreads.
>
>
> I'm perfectly fine with LinuxThreads and I realize that NPTL is a bit new,
> but the problem is that gcc is not suppose to run just on Linux, so a
> general threading model would be better.
>
> Any thoughts?
>
> Biagio
>
>
>
>
> _______________________________________________
> Gomp-discuss mailing list
> address@hidden
> http://mail.nongnu.org/mailman/listinfo/gomp-discuss