I suspect that the sample code in the protocol spec has some bugs :)
The timeout is needed for reliable communications over a modem, where
characters
are often garbled or dropped. Over short cables, it is not usually a
problem. I don't know
of anyone else who implements the timeout. But even in tournaments with
local cables,
sometimes there are errors. I've seen Many Faces retransmit in
tournaments, and I've
seen programs have their links hang.
This is one area where Ethernet would be vastly superior to GMP, since it
has error
correction built in.
Among the thing I do not get in the protocol: should there be a
4 second timeout and retry? David Fotland's code seems to do it,
but I could not find that in William Shubert's code.
Francois Grieu