Re: MPI code error raises a question

Greg Lindahl <lindahl@[EMAIL PROTECTED]
> wrote:

> In article <1igkv0b.w8tcvl1h5dm2yN%nospam@[EMAIL PROTECTED]
>,
> Richard Maine <nospam@[EMAIL PROTECTED]
> wrote:
> 
> >Yes, by the way, there are occasionally reasons to do things like that.
> >They aren't common, but they do come up. And it is even vaguely
possible
> >that MPI might sit in one of the corners where such a thing is useful
> >(in particular, if there is type cheating going on that would cause
> >compilers to ***** if they noticed).
> 
> I'm kind of astonished that you're unaware of the huge problems MPI
> has with the way Fortran modules are specified.

Why would you be astonished? I have never in my life used MPI. Although
I occasionally see discussions about it here and elsewhere, it certainly
isn't something I'm intimately involved in... or have used at all.

I actually have no clue what you are talking about in terms of it having
problems with the way modules are specified. I'm not saying that you are
incorrect in saying that there are such problems. I am simply saying
that I have never heard of them.

If, perhaps, you are talking about problems relating to TKR-cheating,
I've heard about that, but I don't consider that to be a problem with
"the way modules are specified", as it has to do with explicit
interfaces, independent of modules. That's what I was alluding to in the
quote above.

If it isn't that, well... all I can say is that I'm astonished that you
would assume that I have identical interests and expertise to you. Do
you assume that anyone who does Fortran is necessarily into MPI?

-- 
Richard Maine                    | Good judgement comes from experience;
email: last name at domain . net | experience comes from bad judgement.
domain: summertriangle           |  -- Mark Twain