Table of Contents

# $EPIC: exec_command.txt,v 1.3 2006/08/30 21:02:47 sthalik Exp $

Synopsis:

exec [<options>] [<shell commands>]
exec -direct <commands>
exec -name <name> <process id|shell commands>
exec -out <process id|shell commands>
exec -msg <nickname|channel> <process id|shell commands>
exec -notice <nickname|channel> <process id|shell commands>
exec -in <process id> <text to send to process>
exec -window <process id|shell commands>
exec -<signal> <process id>
exec -close <process id>
exec -closein <process id>
exec -line {epic commands} <process id|shell commands>
exec -linepart {epic commands} <process id|shell commands>
exec -error {epic commands} <process id|shell commands>
exec -errorpart {epic commands} <process id|shell commands>
exec -end {epic commands} <process id|shell commands>

Options:

-direct executes commands without spawning a shell
-name give the subprocess a (new) logical name
-out send subprocess output to your current channel or query
-msg send subprocess output to a nick or channel with MSG
-notice send subprocess output to a nick or channel with NOTICE
-in send a text string to the specified subprocess
-window redirect subprocess output to the current window
-<signal> send SIG<signal> to the subprocess.
Numeric signals are also permitted.
-close forcibly kill the subprocess and its children
-closein close the input of the process but let it continue.
-line run this command for each line of output.
-linepart run this command for each partial line of output.
-error run this command for each line of STDERR output.
-errorpart run this command for each partial line of STDERR output.
-end run this command when the process terminates.
See comments on -END below.

Description:

EXEC allows you to spawn other processes from within your client, which are naturally treated as child processes. For example, if you wished to check your email without suspending your client (or quitting irc), you could simply EXEC your favorite mail program.

Processes can be given unique names, allowing them to be referred to more intuitively than trying to figure out its process id number. Processes can also accept any signal normally available through use of the Unix “kill” command (letting you kill, suspend, etc. any process that accepts such a signal). Process identifiers begin with a “%”. Thus, a process named “mail” with pid 0 could be referred to as either “%mail” or “%0”.

EXEC can take several different flags, each of which allow you to manipulate the subprocess in different ways. At the most basic level, the only argument you really need to give is the external program to run. If no arguments are given at all, a list of any currently running processes is returned.

When using the -OUT, -MSG, or -NOTICE flags, output can be sent to the target upon startup of the subprocess, or redirected to it after it's already running by using its process id.

Another flag of note, -CLOSE, can be used to shut down renegade subprocesses that simply won't go away (i.e. with their own quit command, with the KILL signal, etc.). This actually just severs the link between the subprocess and the client, so it will die on its own the next time it tries to send data to the client.

Something to remember when worried about a shell processing arguments is the -DIRECT flag, which executes the commands without spawning a shell. This can cut down on security risks, and save some memory.

The options -LINE, -ERROR, -LINEPART, -ERRORPART and -END all take arguments which must be enclosed in braces “{}” and which are executed for particular kinds of output that the process might give. The $* variable will contain that output when those commands are executed.

Note that the -END flag is different from the other flags. When you specify a %process instead of shell commands, the flags you specify become the new options for the currently running process of that name (if possible), replacing the current settings for those options. Specifying a -END flag in this mode is will add to, rather than replace other -END options.

Examples:

(Assume that the process used with these examples is the Unix mail program.)

To start a process:

    /exec mail

To start a process without a shell interfering:

    /exec -direct mail

To start a process and give it a human-readable name:

    /exec -name mail mail

To send the output of a new subprocess to your current channel or query:

    /exec -out mail

To send the output of a running subprocess to JoeBob with a MSG:

    /exec -msg joebob %mail

To send the output of a new subprocess to #blah with a NOTICE:

    /exec -notice #blah mail

To send a text string (command) to a running subprocess:

    /exec -in %mail ?

To send a running subprocess the KILL signal:

    /exec -kill %mail

To forcibly close down a stubborn subprocess that just won't die:

    /exec -close %mail

To rename a process:

    /exec -name MyMail %mail

To make all mail output bold:

    /exec -line {echo $chr(2)$*} %mail

To add more -end commands after the process has started:

    /exec -end {commands} %mail

Which is equivalent to:

    /wait %process -cmd {commands}

Aliases:

When using EXEC with the -IN flag, the functionality is identical to using MSG to send a message to a process. See MSG for more information.

When using the -END flag, the functionality is identical to using “WAIT %process -cmd {command}”.

Restrictions:

Use of external programs from within irc can sometimes be dangerous if you don't know what you're doing. The danger doesn't necessarily lie in the external programs themselves, but rather in any lack of knowledge about them you may have. When all else fails, don't use a command if you don't know what it does.

You can explicitly limit your ability to use EXEC in several ways. If set SECURITY has the 1 bit set, the client will only permit EXEC to be used interactively; using it from within aliases is not allowed. For real security, you can #define RESTRICTED at compile time, which will completely disable EXEC.

Other Notes:

The available signals that may be sent to a process will vary from system to system.