Hi,
> Apart from typos etc. it seems reasonably ok - if somewhat complicated.
> (It's sufficiently complicated and opaque that I'm not sure exactly what
> effect you're trying to achieve here.)
Indeed! I will stop here, since it is obvious it is completely unclear what
I am saying. These codes of mine is ficticious, so if you look at them
without knowing what it is I am trying to do it seems like they can be
rewritten trivially. For example,
CALL test(handle,proc)
*can not* be replaced with
CALL proc(handle)
because there are other things going on in test than just calling proc
(these are not shown and I did not put "..." to keep the code shorter).
Similarly, making test type-bound as in your example only slightly
simplifies the syntax, but not really any content.
In my last (likely vain, but hey :) attempt, I will resend the code for a
reverse-communication based conjugate-gradient code from last Friday
corrected to reflect all the mistakes I made and that you and Kurt have
corrected. But it is a long piece of code (albeit coherent and complete, so
relatively easy to understand). I should be done in an hour or so. We have
discussed (and you have before I joined the list) reverse-communication
before, but this is a rather "different" solution to it, so fear not
threading of old water...
Off to work...
Aleksandar
_____________________________________________
Aleksandar Donev
http://www.pa.msu.edu/~donev/
[log in to unmask]
(517) 432-6770
Department of Physics and Astronomy
Michigan State University
East Lansing, MI 48824-1116
_____________________________________________
|