The runat utility is used to execute shell commands in a file's hidden attribute directory. Effectively, this utility changes the current working directory to be the hidden attribute directory associated with the file argument and then executes the specified command in the bourne shell (/bin/sh). If no command argument is provided, an interactive shell is spawned. The environment variable $SHELL defines the shell to be spawned. If this variable is undefined, the default shell, /bin/sh, is used.

The file argument can be any file, including a directory, that can support extended attributes. It is not necessary that this file have any attributes (or be prepared in any way) before invoking the runat command.

The following operands are supported:

file

Any file, including a directory, that can support extended attributes.

command

The command to be executed in an attribute directory.

A non-zero exit status will be returned if runat cannot access the file argument, or the file argument does not support extended attributes.

See fsattr(5) for a detailed description of extended file attributes.

The process context created by the runat command has its current working directory set to the hidden directory containing the file's extended attributes. The parent of this directory (the ".." entry) always refers to the file provided on the command line. As such, it may not be a directory. Therefore, commands (such as pwd) that depend upon the parent entry being well-formed (that is, refering to a directory) may fail.

In the absence of the command argument, runat will spawn a new interactive shell with its current working directory set to be the provided file's hidden attribute directory. Notice that some shells (such as zsh and tcsh) are not well behaved when the directory parent is not a directory, as described above. These shells should not be used with runat.

example% runat file.1 ls -l
example% runat file.1 ls

example% runat file.2 cp /tmp/attrdata attr.1
example% runat file.2 cat /tmp/attrdata > attr.1

example% runat file.2 cat attr.1 | runat file.1 "cat > attr.1"

example% runat file.3 /bin/sh

The above list of commands are known to work with runat. While many other commands may work, there is no guarantee that any beyond this list will work. Any command that relies on being able to determine its current working directory is likely to fail. Examples of such commands follow:

example% runat file.1 man runat
getcwd: Not a directory

example% runat file.3 /usr/bin/tcsh
tcsh: Not a directory
tcsh: Trying to start from "/home/user"

example% /bin/pwd
/

SHELL

Specifies the command shell to be invoked by runat.

The following exit values are returned:

125

The attribute directory of the file referenced by the file argument cannot be accessed.

126

The exec of the provided command argument failed.

Otherwise, the exit status returned is the exit status of the shell invoked to execute the provided command.

See attributes(5) for descriptions of the following attributes:

open(2), attributes(5), fsattr(5)

It is not always obvious why a command fails in runat when it is unable to determine the current working directory. The errors resulting can be confusing and ambiguous (see the tcsh and zsh examples above).