Out of curiosity, does anyone know roughly what percentage of players use MCCP when the server supports it?
Edit: I'm, of course, not looking for hard numbers. A wild guess will do fine. I'm just wondering if it's worth implementing MCCP in a from-scratch codebase.
Probably a large number of them do. Most people are using either zMud or MUSHClient and I believe they both activate MCCP by default. So it would be worth the time to include the support in your codebase.
02 Jun, 2008, David Haley wrote in the 3rd comment:
Votes: 0
Don't most MCCP implementations support telnet negotiation for it, meaning that if the client supports both MCCP and negotation (and both MUSHclient and zMud do, I think) it will get turned on automatically anyhow?
I don't know about the negotiation aspects, but I do know that on my mud my statistics show that it's saving me ~3x the bandwidth that my mud actually ends up otherwise expending, so I'd definitely say it's worth including.
If the codebase supports telnet negotiation properly, then yes. In theory that would do the job. But that also requires the client program to support it properly as well, and the combinations aren't always there. Add the support to the codebase and let the client authors sort out their own bugs :)
Well, cutting bandwidth usage to 33% of non-compressed bandwidth seems sort of good. On the other hand, the amount of bandwidth a MUD uses is fairly trivial, anyway. :-)
I think I will be implementing MCCP, but not in the first version of the server. I want to get "something" up and running before I start worrying about adding chrome.
Thanks!
Edit: BTW, I just noticed… CCCP is the abbreviation in Russian for "Union of Soviet Socialist Republics." MCCP == CCCP? Coincidence? Ok, Ok, I'm just being silly for the sake of being silly now. :P
02 Jun, 2008, David Haley wrote in the 7th comment:
Votes: 0
That's probably a good plan. Fortunately, MCCP is rather easy to add on later: presumably, your code will funnel everything for the network into one function anyhow, and you'll just need to stick a layer in between that compresses it if desired.
I've said this before and I still agree. MCCP is a hindrance unless you host your mud yourself. It uses a substantial amount of RAM and bandwidth (at least not the minuscule amount a mud uses) isn't an issue nowadays on most servers.
07 Jun, 2008, David Haley wrote in the 13th comment:
Votes: 0
MCCP uses a substantial amount of RAM? As they say: "o rly?" Processor, maybe (and even then in small amounts), but not RAM…
23 Jun, 2008, MaineCoon wrote in the 17th comment:
Votes: 0
Here is my experience with MCCP, since installing it a year ago.
Use of MCCP is automatic - at login, telnet negotiation offers it, and if client accepts, then it is enabled. There is no way to opt out short of forcing it off at the client.
RAM usage - investigate when initially installed a year or so ago. Saw an increase of about 8 megs of RAM when first active, for 25 players online.
CPU usage - again last measured when initially installed, and this was on our old host. CPU usage increased by about 10-20% relative. This was about 1% extra cpu usage of a Pentium 3 Xeon 1ghz.
% of users: As of this moment, out of 23 users, 17 have compression active.
Here is a capture of data from my MUD after 18 hours uptime today:
38 megs of data sent, 66.5 megs of bandwidth usage saved due to compression. So about 2/3s bandwidth saved.
Bandwidth usage for me for a month is about 5 gigs for the MUD, with MCCP in use. Since my virtual host gives me 200 gigs bandwidth, increasing from 5 to 15 wouldn't hurt. On the other hand, some of our players who tend to have laggier connections have noticed an improvement in latency, which is important for some of the fast paced PK.
25 Jun, 2008, darkraider wrote in the 18th comment:
Votes: 0
What I want to do is give players a bonus if they have mccp enabled.
Edit: I'm, of course, not looking for hard numbers. A wild guess will do fine. I'm just wondering if it's worth implementing MCCP in a from-scratch codebase.