[MUD-Dev] Re: Question regarding Java threads

[Chris Gray]

[Vadim Tkachenko:]

 >> Character-mode telnet connections are indeed evil. They consume far
 >> more resources than a line-mode telnet connection.
 >
 >What is clear, is that the line mode allows buffering and utilizes the
 >protocol features better than character mode. Also, the interim solution
 >is WON'T ECHO, DO ECHO, if I remember it right, so the echo wouldn't
 >travel back and forth. Is there anything else which is a reason for such
 >evilness?

The order of magnitude increase in packets is my main concern. The
higher typing latency for the user is another. Also, as I believed I
mentioned in a previous note, my experience with trying to use
DO ECHO and DONT ECHO to toggle the client's echoing of passwords
just doesn't work in many telnet clients.

 >> Of course the full solution is for everyone to use my not-yet-written
 >> Java client, which connects in a binary mode which doesn't even have
 >> the concept of a character mode. :-/
 >
 >Hold on a second, I might forget some details, but isn't it right that a
 >character mode _IS_ a binary mode with the command sequences in between?

Depends on your definitions, I guess. I look at character mode as just
being an 8 bit character mode that has a few escape conventions. A true
binary mode doesn't need any escapes since each packet has a binary
header that contains any needed control information. So, a binary
mode transmits slightly more data, but it always does so as larger packets.

--
Chris Gray     cg at ami-cg.GraySage.Edmonton.AB.CA