Network and communications questions

[Dar Scott]

On Wednesday, February 4, 2004, at 05:58 PM, Brent Anderson wrote:

> I am writing a group of programs designed to communicate with each 
> other over a network and wanted to know specifics on the best way to 
> have one program send and receive over a network connection, how to 
> set it up in Revolution, and any other tips and commands that I can't 
> seem to find in the documentation. Another question I have is how 
> Revolution communicates over a network in the first place (does it use 
> peer to peer or does it need a computer to act as a server, or some 
> other form of connection?)

At the sockets level (low), you do not need to have a server, but you 
might want to design around one, especially if firewalls might be 
involved.  This level is TCP/IP UDP and TCP.  UDP is an "unreliable" 
message protocol.  TCP is a connection oriented "reliable" byte-stream 
protocol.

I have a very simple remote send over UDP library that will work for 
small commands at a low rate.  The command protocol you set up should 
send confirmations.  If sockets is too low level, I can send you that, 
if I can find it.  Most of the source code consists of caveats in the 
comments.

However, at the higher levels you might use an http server and have all 
apps communicate with that.  I know there is some level of SOAP 
available.  I think Andre Garzia has tinkered with an alternate RPC 
protocol over http.

I don't know the status of libIPC; Jan could say better.

Alex Rice has a Rendezvous external should you need it in getting into 
finding services.

If this is a simple network communications requirement, I'd start with 
UDP sockets, datagrams.  Look at Transcript dictionary entries for 
'open socket' 'close socket' 'write to socket' 'accept' 'read from 
socket' and related messages.

Dar Scott