usepackage - Online in the Cloud

PROGRAM:

NAME

usepackage - Usepackage Environment Manager

SYNOPSIS

See use(1).

DESCRIPTION

Usepackage is an environment management program. It is based on the principle of packages
- collections of executables that share a common set of necessary environment variables,
such as PATH, MANPATH or LD_LIBRARY_PATH.

For each given package, use sources the appropriate environment information into the
current shell. The environment information is specified in packages files, see PACKAGES.

The usepackage executable is the backend program used to generate environment information
suitable for sourcing into a running shell. The use frontend should be invoked to actually
effect changes to the environment, see use(1).

OPTIONS
-v Output verbose information to the standard error stream.

-s Silence warnings for un-matched packages. This is useful in a shell rc script when
a package is known not to be available on all architectures that the shell is used
on, see PACKAGES.

-c Force csh style environment output.

-b Force bourne shell style environment output.

-f file
Specify an alternate initial configuration file, see PACKAGES.

-l List available packages and groups, see ANNOTATIONS and GROUPS.

PACKAGES

Usepackage reads package environment information from the configuration file as follows:

LOAD ORDER
When Usepackage starts up, it loads an initial configuration file, the name of this file
may be given on the command line (see OPTIONS), otherwise it defaults to the builtin name
"usepackage.conf" (see FILES). This file is located by searching along a path which
defaults to:

/usr/etc:~:.

This path may be overridden with the PACKAGES_PATH environment variable (see ENVIRONMENT).
A packages file may contain inclusion directives which cause the sourcing of other files
at given points. An inclusion directive looks like:

(include file-name)

The same location mechanism is used to find "file-name" as for the initial configuration
file. A package file name may also be given as an absolute file name or may be shell-style
user directory, tilde (~), relative.

SYNTAX
A package file consists of comments, delimited by a leading hash (#), or package
definitions of the form:

package [arch [os [version [host [shell]]]]]
[<= requires ...] : setting [, setting ...] ;

The package, arch, os, version, host, and shell parts may be simple shell-style patterns
of the form:

* matches anything.

foo* matches "foo", "foobar" etc.

{foo,bar} matches "foo" or "bar".

The requires list specifies the names of other packages which must be sourced into the
environment before the settings for this package are processed.

A setting is either a variable definition (as described below) or a section of text to be
directly evaluated in the shell, delimited with <[ and ]>, for example:

<[ /usr/local/bin/test-init ]>

When Usepackage searches for the definition for a particular package, it compares each
line in the packages file against the name of the package given (package) and system-
dependant information for the execution host, as obtained by uname(2). This information
is the hardware implementation (platform), the operating system name (os), the operating
system version (version) and the hostname (host). Comparisons are case-insensitive. If a
match is obtained then the given variable definitions and script sections are processed to
modify the environment. A variable definition may have one of the following forms:

var-name = "string"
var-name = path-list
var-name += path-list
var-name += "string"

The first sets the given variable to a literal string value, the second sets the given
variable to a path list, the third prepends the current value of a variable with the given
path list, and the fourth interprets the literal string as a path list and prepends it to
the current value of the variable. Path lists are colon (:) separated lists of directories
and may contain shell-style tilde (~), user-relative, directories which will be expanded
automatically (except when the path list is given as a literal string). When pre-pending
paths to a variable, duplicate paths are removed from the original value first.

In addition, Usepackage can optionally test for the existence of paths before setting or
adding them to variables. This is done with the test form of the above operators:

var-name ?= path-list
var-name ?+= path-list
var-name ?+= "string"

These operate as per the definitions above, but will ignore path components that don't
exist.

GROUPS
In addition to the package definitions in a packages file, there may also be group
definitions. These have the following syntax:

group := package [, package ...] ;

Usepackage searches for a given package name in the defined groups first, if the given
name matchs a group name then the packages defined as part of that group are sourced into
the environment together. A group definition may not reference other groups and may not
contain patterns.

ANNOTATIONS
In order to give useful package information to the user, annotations may be placed in the
packages file that give summaries of packages. These annotations have the form:

>> name : "description" <<

Usepackage collects these annotations together and displays them when called with the -l
flag, see OPTIONS. These annotations have no impact on the package mechanism and need not
necessarily be beside or correspond to the package definitions (although this is the
sensible way to arrange things).

EXAMPLE
The following fragment of a packages file illustrates the main features:

# GNU software is available everywhere:

>> GNU : "The GNU project software" <<

GNU : PATH += /usr/local/gnu/bin,
MANPATH += /usr/local/gnu/man ;

# CVS requires RCS which is found in the GNU package, but
# is only available on SPARC Solaris machines:

>> CVS : "Concurrent Versions System revision control" <<

CVS sun4* SunOS 5.* <= GNU :
CVSROOT = /usr/src/cvsroot,
CVSEDITOR = "vi",
PATH += /usr/local/cvs/bin,
MANPATH += /usr/local/cvs/man ;

# User bin directories (Solaris will run SunOS 4 binaries):
# (Only added to the path if the directories actually exist)

>> user : "User's own programs" <<

user sun4* SunOS : PATH ?+= ~/bin/sun4 ;
user sun4* SunOS 5.* : PATH ?+= ~/bin/solaris ;
user alpha OSF : PATH ?+= ~/bin/alpha ;

# Special function for zsh:

>> zsh-function : "Special zsh function" <<

zsh-function * * * * zsh :
<[ hello() { echo "Hello World!"; } ]> ;

# include standard packages:

(include packages.standard)

# security hole:

dot : PATH += . ;

# groups:

user-setup := standard, user ;
programmer-setup := standard, CVS, user, dot ;

Note the use of Operating System version numbers to distinguish between SunOS 4 and
Solaris (SunOS 5), the use of "sun4*" to match the multiple different platform versions of
SPARC machines (sun4m, sun4c, etc.) and the fact that package "user" on a SPARC Solaris
machine will match both of the first two lines of the "user" package section, resulting in
the "solaris" directory and the "sun4" directory being added into the PATH.

Assuming the appropriate shell setup script has been sourced - see use(1) - then the
following command will cause the CVS environment (including the GNU environment) to be
sourced:

$ use CVS

Note that on a DEC Alpha machine, this will generate a warning like:

$ use CVS
warning: no match for package `CVS' on this host.

In a shell script which is executed on a number of different platforms (such as the shell
startup script). These warnings may be silenced (see OPTIONS).

Use usepackage online using onworks.net services