"Martin James" <mjames_falcon@dial.pipex.com> wrote in message
news:43817d1c@newsgroups.borland.com...
>
> If you possibly can, stuff the timeout - try to design so it does not
> matter.
--------
Thanks for your response. You're quite right. I must just accept the
'problem' and design to accomodate it without affecting performance. This
can be done with threading I believe (green as grass here I'm afraid) , so
I am currently looking into that. I've come across some good articles on the
subject and I think using the BeginThread function I can do it.
I'll fire up a few threads and if one gets stuck (doesn't respond with x
seconds) I'll fire up another one and let the old one conk out to it's
hearts content. Once it does come back to life I can dump its now out of
date info and return it to the general thread pool.
(you get some really novel and catastrophic problems playing with threads -
my whole machine dissapeared off into the wilderness as my, not working,
attempt hogged everything)
