Whether one likes the bottom-level details of Craig's socket re-do or not (
and I'm a little nervous about needing to make threads work on Acorn) the
fact is that the currnet socket stuff needs a lot of cleaning to handle
timeouts, closed sockets etc better. 'Halt'ing to warn about a timeout is
less than useful when you can't proceed without getting later uglier
faults. With better  socket code, easier streaming and excpetions raised
correctly, I strongly urge a move to something at least basedon Craig's
stuff.
I would point out that it was originally written for the Interval
RTOS stuff we were working on, and got tested _extremely_ thoroughly on
three platforms (including our ST-RTOS on raw hardware) by a dozen people
for six+ months. That's better testing than many commercial products get
:-)

Even though I stand to 'suffer' most, I think it makes sense.

tim


-- 
Useful random insult:- If she was any dumber, she'd be a green plant.
Tim Rowledge:  rowledge at interval.com (w)  +1 (650) 842-6110 (w)
 tim at sumeru.stanford.edu (h)  <http://sumeru.stanford.edu/tim>