zoiduser - Extended user documentation for zoid



The location of config and data files can be installation specific. You can check the location where zoid looks for them with the command zoid --config.

By default zoid uses two run control files: /etc/zoidrc and ~/.zoidrc or ~/.zoid/zoidrc, these are ordinary perl scripts that can interface with zoid. In general data files are found in ~/.zoid/, /usr/local/share/zoid/ or /usr/share/zoid/ these are called 'data_dirs'.

The sub dir plugins can contains config files for various plugins or directories with all files belonging to a plugin. The following plugins are included in the standard distribution:

FIXME tell which builtins are provided by each plugin FIXME tell commands provided by


Ships a collection of standard builtin commands that one would expects to have in a shell.

Provides: cd, pwd, exec, eval, source, true, false, newgrp, umask, read, wait, fg, bg, kill, jobs, set, export, setenv, unsetenv, alias, unalias, dirs, popd, pushd and symbols


Keeps a history of your doings.


Takes care of tab expansion.

Provides: complete


Tries to load a module from the Term::ReadLine family of modules.

Provides: readline, readmore and select


Adds a 'CPAN mode' which uses CPAN::Shell.

In theory you should be able to remove or replace any of these plugins, but you better not try this till a more stable version of zoid.

Key Bindings

Key bindings depend on the ReadLine module that is loaded. See for example Term::ReadLine::Zoid.


The following settings supported by (some) Zoidberg modules. They are housed in the hash ->{settings}. Also the builtin set command acts on this hash.

( FIXME link to doc on set command )

More advanced settings can be found in zoiddevel.


Disable path expansion for shell-like syntax.


When set makes globs case insensitive.


It allows to have wildcard patterns that expand into an empty list. The default behavior is to leave the pattern unchanged when it is syntactically incorrect, or when the list of matching pathnames is empty.


Do brace expansion posix compliant instead of bash compatible. This means that braces will be removed even if there is just one word between them.


Hide all object methods and data structures which have a name starting with an '_' unless they are asked for explicitly.


Hide all filesystem nodes which have a name starting with a '.' unless they are asked for explicitly.


When this setting is in effect a ^D char, also known as EOF or EOT, won't exit the shell.


Don't hide the root object behind clothes for things like tab completion and the single arrow operator, but show it in full glory.


Asynchronous notification, don't wait for the respawning of a prompt before notifying the user about background jobs that have finished.

This option should be named 'notify_async' for clearity, but 'notify' is the proper name according to POSIX.


When you stop a job the shell will show all jobs instead of just notifying just of the one you stopped; makes the behaviour more tcsh(1) like.


Since the syntax of the Zoidberg shell is completely configurable we can only comment on the syntax as defined by the default config files. The general structure will be the same for most user defined configuration but any markup or token could be changed to render the syntax completely unrecognizable.

The Zoidberg syntax consists of three levels of grouping:

Logical grouping

First the syntax is split into blocks with logical/script delimiters

        [block] && [block] || [block] ; [block]
Pipes and redirections

Once the string is cut into logical blocks each of these blocks is split in sub blocks by pipes and.

        [sub_block] | [sub_block] | [sub_block]

The exit status of a pipeline is the exit status of the last command in the pipeline.

Context blocks

At last for each of these sub-blocks a context is decided like:

        CMD     - commands sytax (like sh syntax)
        PERL    - blocks of perl code

Each is executed differently by a suitable subroutine or application, and all are glued together to form a pipeline.


The contexts named below are hardcoded, others can be added by plugins. To disable hardcoded contexts see the _no_hardcoded_context setting.


Perl is the default context if the whole block is between curly brackets, when the first non-whitespace char of the block matches $,@ or % , when the first word seems to be a subroutine because it has parenthesis or when the first word is a perl keyword like 'if', 'while', 'for' etc. These keywords can be configured with the array in ->{settings}{perl}{keywords}

        # perl because of reserved word
        zoid$ for (0..3) { sleep 1; print $_ . "\n" }
        # perl because of dollar sign
        zoid$ $self->{settings}{naked_zoid}++
        # perl because of parenthesis
        zoid$ ls(qw/-al/)
        # perl because of curlies
        zoid$ { open TEST, '<test.dat' }

Perl code can have modifiers after the last curly. The default can be set with ->{settings}{perl}{opts}. Currently supported are :

        n: enclose the expression in a "while STDIN" loop
        p: like 'n' but also print $_ at the end of each loop
        g: grep lines from STDIN that make the expression return non-zero
        z: use strict
        Z: no strict, the default
        # example
        zoid$ ls -al | { s/^(d)\S+/DIR:/ }g

Zoidberg applies a bit of source filtering is applied to the perl code. This renders a lonesome dereference operator -> into $self->.

        # Thus
        zoid$ ->kill('1230')
        # is the same as
        zoid$ $self->kill('1230')

Variables in that match the name of an environment variable get translated to the environment varriable, arrays cause a tied array to be imported from Env. You can manipulate the existence of an environment variable with the 'export' builtin.

        # Thus
        zoid$ print $PATH
        # is (if $ENV{PATH} exists) the same as
        zoid$ print $ENV{PATH}
        # and this also works
        zoid$ push @PATH, '/usr/X11R6/bin/'

To avoid source filtering use the 'z' option. ( Although once a tied array is imported, it's there. )

Non-existent sub routines are AUTOLOAD'ed to be built-in or system commands, alias- and other expansions apply.

        # Thus
        zoid$ ls('*')
        # is perl, but does exactly the same as
        zoid$ ls *

When such a AUTOLOAD'ed routine is used in scalar or list context it's output will be captured and returned as scalar or list.

        # examples
        zoid$ ls($_) for cat('MANIFEST')
        zoid$ mplayer( locate('blinkenlights.mpg') )

For this "capturing mode" the record separator ($/) can be set by using $ENV{RS}.


This context is intended to make the Zoidberg shell a little friendlier to people used to shells like bash(1). Also this syntax requires less chars to execute a system or built-in command. Only the most basic stuff is implemented, you should use perl for things like flow control. Since the default syntax for pipelines and logic lists is also the same as in "sh like" shells a lot of simple constructs will work as expected.

        # this does what it would in /.*sh/
        zoid$ ls -al | grep -v CVS | grep ^d > dirs.txt

Quoting does not entirely work like in "sh like" shells. Most importantly, you can't start a quoted block in the middle of a word, if you use quotes you need to quote an entire word. (The exception to this rule is syntax like export var='val'.) Also be aware that within single quotes the backslash is still recognised as an escape, but only for the quote itself, the same way perl does.

All words, are subject to parameter- and path-expansion unless they are quoted. For double quoted words only parameter expansion is performed, for single quoted words no expansion is performed at all. Parameter expansion interpolates environment variables. (Path expansion is sometimes also called 'glob expansion')

All parameters can be used as arrays, the colon ':' will be used as element separator. To get one element the form $VAR[$i] is recognized by the parameter expansion, where $i is a null-based integer. The form @VAR is a special case that will be expanded to a word list of all the array elements. To avoid conflicts with things like the 'user@host' format used by many applications this special case can only be used as a separate unquoted word.

        zoid$ echo $PATH
        zoid$ echo $PATH[2]
        zoid$ echo @PATH
        /bin /usr/bin/ /usr/local/bin

If the command matches an entry in the $$shell{commands} hash, it is taken to be a built-in command and the job is not forked to a seperate process.


FIXME comment on custom defined contexts

FIXME comment on aliases


Like most shells zoid mixes the use of it's own variables with the use of environment variables. You can use the export builtin to move variables from the perl namespace to the environment.


While in normal use zoid tries to determine which context a statement belongs to automaticly there are also situations in which you want to force one context for all commands. In these cases you can set the 'mode' of the shell. The "mode" builtin is provided to do this a little more intelligent.

For example the CPAN plugin provides an interactive interface to the CPAN::Shell module.

        # enter cpan mode
        zoid$ mode cpan
        zoid$ i /Zoidberg/
        # use a "normal" shell command
        zoid$ !df -h /
        zoid$ install Bundle::Zoidberg
        # return to the "normal" mode
        zoid$ mode -

Modes can be either the name of a custom context (probably defined by a plugin), or the name of a perl module (matching a '::'). If the mode is a module name all commands are mapped to methods in that namespace; in this case you can postfix the module name with '->' if the module expects the class name as first argument to all calls.

        # set mode the hard way
        zoid$ mode CPAN::Shell->
        zoid$ i /Zoidberg/
        zoid$ mode -


Some system command:

        zoid$ mplayer -vo sdl -ao sdl my_favorite_movie.avi

Perl code between brackets:

        zoid$ { print "This is perl code." }

A subroutine call to an object loaded in Zoidberg, perl code with a little source filtering:

        zoid$ ->Help->help
        zoid$ ->Help->help('objects')

A builtin command:

        zoid$ cd ..

A custom syntax (SQL) enforced on a block:

        zoid$ sql{ SELECT * FROM users WHERE clue > 0 }

And as said all of these can be combined:

        zoid$ cd /usr/local && { print "This is perl code." } | less || sql{SELECT * FROM users WHERE clue > 0}

This will first execute cd /usr/local, on succes followed by { print "This is perl code." } | less and if one of these failed we get to see the result of sql{SELECT * FROM users WHERE clue > 0}

This makes the precedence of this example as follows.

        ( ( 1 and ( 2 pipe 3 ) ) or 4 )

        1 = cd /usr/local
        2 = { print "This is perl code." }
        3 = less 
        4 = sql{SELECT * FROM users WHERE clue > 0}


(random notes that probably belong elsewhere in this document)

Files starting with a '-' are ignored in globs unless the glob starts with a '-', this is to prevent globs from accidentally adding switches to a command.


The variables $PWD, $HOME and $USER are set to default values if not yet set by the parent process.

The variable $ZOID will point to the location of the zoid executable, it is similar to $SHELL for POSIX compliant shells. zoid uses a differ- ent variable because some programs seem to expect $SHELL to point to a POSIX compliant shell.

To switch off ansi colours on the terminal set $CLICOLOR to 0 (null).

The variable $RS (record seperator) is used to split lines when capturing ouput from a command in list context. It defaults to the newline character.

The variable $IFS (input field seperators) is used to split words after several expansions. It defaults to <space><tab><return>.

This list probably isn't complete.


Zoidberg uses rc files, data files and plugin files, use the --config switch to check the search paths used.

Which rcfiles are loaded is controlled be the 'rcfiles' and 'norc' set- tings, try "zoid -o norc" to skip all rcfiles or "zoid -o rcfiles=file1:file2:file3" to use files other then the default.

The runtime search path for plugins etc. can be controlled with the 'data_dirs' setting, try "zoid -o data_dirs=dir1:dir2:dir3".


zoidbuiltins(1), zoiddevel(1), zoidfaq(1), perl(1),