CTCP and DCC FAQ

Revision as of 19:21, 3 August 2007

What does CTCP stand for?

CTCP stands for Client To Client Protocol. This is not to be confused with the CCCP ;)

Why is it called Client to Client Protocol?

Because it is sent and interpreted by your IRC client. The IRC server has nothing to do with CTCP. It treats a CTCP message as just another /msg to someone. Unlike other commands, for example /whois.

How does CTCP work?

In CTCP, there are two forms of communication: COMMAND and REPLY. Commands are sent as follows:

PRIVMSG ((target):^A ((command>) ((value)^A

Where - ^A = ASCII Character 1
- command = CTCP command
- value = value to be processed by the target's client

And REPLIES in this format:

NOTICE ((target):^A ((command) ((value)^A

Where - ^A = ASCII Character 1
- command = CTCP command
- value = value to be processed by the target's client

For example, a ping of channel #chatx would be:

PRIVMSG #chatx :^APING 866780265^A

This message would be received by all the people on channel #chatx but yourself. They would all get this:

:Joe!duck@opera.iinet.net.au PRIVMSG #chatx :^APING 866780265^A

For example, a reply to this ping would be:

NOTICE Joe :^APING 866780265^A

This message would be received by Joe like this:

:Skinner!sjm076@ppp-per-217.ca.com.au NOTICE Joe :^APING 866780265^A

Joe's client would then interpret this as being a reply to a PING and process it accordingly, in this case, taking the timestamp (866780265) and subtracting it from the current timestamp.

What are the common CTCP commands?

CLIENTINFO, ACTION, DCC, ERRMSG, FACE, FINGER, PING, SOUND, SOURCE, TIME, USERINFO, VERSION, and XDCC.

Briefly, this is what each command does:

CLIENTINFO -- Display valid CTCP commands for that client.

ACTION -- when you type /me does this, it is actually sent as a CTCP message. However, unlike other CTCP commands, it does not require a REPLY, and none should be given.

DCC -- DCC is established by CTCP, however it is not conducted over CTCP. See the section on DCC for more information on this protocol.

ERRMSG -- Reply to unknown CTCP command.

FACE -- Pictographic 32x32 pixel representation of user (Macintosh clients only)

FINGER -- Shows idletime of client, usually with an e-mail address and message.

PING -- Used to measure lag, or the time it takes for information to travel between servers.

SOUND -- Causes the client(s) that receive it to play a sound file.

SOURCE -- A URL where the client/script can be obtained.

TIME -- Local time of recipient.

USERINFO -- Shows a "witty" saying set by user.

VERSION -- Replies with IRC client version.

Why don't some IRC clients support all these?

Because they don't basically ;) However, nearly every IRC client will support CLIENTINFO, which shows what commands that client supports.

How do I send a CTCP command?

"/CTCP ((target) ((command)" is implemented in nearly every IRC client. For those that it isn't, usually commands like /version ((target), /ping ((target), etc., are. The exception to this is PING, which is usually done by /ping ((target) (or /cping ((target) in ircle).

It should be noted that /version is an ircd command as well.

What is a CTCP flood?

It is when someone maliciously makes your client reply to CTCP messages with the purpose of making you disconnect from the IRC server.

How do CTCP floods work?

Most servers are set up to allow you receive more data than you can send; therefore some commands such as CLIENTINFO can be used to disconnect you.

The command to make your client reply to a CLIENTINFO is 12 characters (plus source/target information), while the reply to CLIENTINFO may be 100 - 200 characters. You will be disconnected with the message "Excess Flood."

What about DCC?

DCC is started by a CTCP command. However, once it is started, it is conducted totally independant of the IRC server. For example, to start a DCC chat connection, the originator of the request (whoever typed /dcc chat ((nick)] will send something like this:

PRIVMSG ((target) :^ADCC CHAT ((type) ((longip) ((port)^A

^A = ASCII Character 1
type = Either Chat or Talk, but almost always Chat these days
longip = 32-bit Internet address of originator's machine
port = Port on which the originator is waitng for a DCC chat

The person I want to DCC chat would get this:

:Joe!duck@opera.iinet.net.au PRIVMSG SomeOne :^ADCC CHAT CHAT 3406736986 2094^A

From this, their IRC client would know that the machine with address 3406736986 is ready to accept a DCC chat connection on port 2094. The IRC client of the target of the DCC chat would then establish a connection to port 2094 of machine 3406736986 if the DCC chat was accepted.

DCC file sends are similar to DCC chat requests, but follow the format below:

PRIVMSG ((target) :^ADCC SEND ((filename) ((longip) ((port) ((filesize)^A

Where - ^A = ASCII Character 1
- filename = Name of file being sent
- longip = 32-bit Internet address of originator's machine
- port = Port on which the originator is waiitng for a DCC chat
- filesize = Size of file being sent

The person I want to send a file to would get this:

:Joe!duck@opera.iinet.net.au PRIVMSG Someone :^ADCC SEND CTCP_and_DCC 3406736986 2097 4509^A

From this, their IRC client would know that the machine with address 3406736986 is ready to send a file named "CTCP_and_DCC" that is 4509 bytes long on port 2097. The IRC client of the target of the DCC send would then establish a connection to port 2097 of machine 3406736986 if the DCC send was accepted.

What can I do if I want more information on CTCP?

• CTCP Specification, written in 1994.
• Unofficial updated draft of the CTCP Spec, adopted by most major IRC client including mIRC.
• DCC Specification

Summary

CTCP commands are sent as privmsgs. CTCP replies are sent as notices.

CTCP commands and replies are in the format of ((privmsg/notice) ((target):^A((command) ((arguments)^A.