On Wed, Aug 11, 2004 at 08:48:54PM +0100, Nick Wedd wrote:
Those who have advocated GTP as a replacement for GMP have missed the
point that it is a different-shaped animal.
I think we have a classical example of a culture clash here. Windows
users assume that all computers have serial ports and will work together
just by sticking a suitable cable between two computers. Unix users
assume that all computers are part of a network, and any communications
is easiest done using that network, using standard protocols.
I agree with your main point here. I will comment only on one detail -
Windows users are nowadays disconcerted to learn that serial ports are
no longer being supplied as standard.In practice, both approaches require some additional tweaking (a serial
port can be something else than what a Windows user expects, and require
a crossed/non-crossed/whatever cable and flow-control in software,
hardware, or something else, plus an agreement on speed, data bits,
parity.... The network has to be there too, with given addresses, ports,
cabling, and everything).
I have been assuming a working Windows network. The sponsors are
(unfortunately) more likely to provide that than a working Unix network.Still, the GTP discussion *is* about two separate things: What the
protocol says about the communications, and how that communication
channel is to be established.
I tend to agree with those who say that the standard does not have to
specify the channel, but that a tournament will have to specify that it
is to use GTP, over TCP/IP, running on (say) a regular 100Mbit
twisted-pair Ethernet etc, where the participating programs initiate the
connection to a server at ...
- Heikki
glad I don't have to organize a tournament...