bash-env

Hierarchical, multi-workgroup, multi-system, distributable bash environment.

Usage:

~/.bashrc:

source ~/.environment.sh

If these bash-env files are kept in another directory, the location needs to be named in the PATH environment variable. For example, if these files are stored in ~/lib (aka $HOME/lib), then the PATH environment variable should be set:

export PATH=$PATH:$HOME/lib

Description:

This script is meant to be invoked from ~/.bashrc or ~/.bash_login/

First, the common platform and host-independent environment setups are done, then platform-specific and then host-specific environment files, if any, are sourced.

The list below represents the general order of the files being sourced:

.environment.sh
.environment.envars.sh
.environment.x11.sh		  # special case
.environment.$OSTYPE.sh
.environment.$OSTYPE-$OSREV.sh
.environment.$MACHTYPE.sh
.environment.$HOST.sh
.environment.$USER.sh
.environment.$WORKGROUP.sh
.environment.$DOMAIN.sh	  # for each name in $DOMAINS
.environment.term.sh	  # special case

By performing the more general files first, followed by the more specific files, exact control, with reasonable defaults can be achieved without duplicating setup code. By having the user file last, the user can override anything to suit taste.

Let's describe some uses.

I have computers at both work and home. There are some production systems on which I have accounts, some QA hosts, some development systems, including my work laptop. My home systems include a couple of Mac OS X systems, an Unbuntu system, and a Mac OS X laptop.

I would like to have as uniform an environment as possible, but each system, or each group of systems will have some unique environments.

The following is a hierarchical set of environments, centrally managed, distributed across all the hosts. Even though some of the environment files are not needed on some hosts, it is easier to manage if all the files are distributed to all the hosts on which I have access.

First, I'll setup two "workgroup" environemnts, one for work, and one for home. My "home" workgroup is called "Home". My work workgroup is a hypothetical company called "Acme.com", and we'll call the workgroup "acme".

The group of work production systems live in the subdomain of "prod.acme.com", the QA systems in "qa.acme.com", and the dev systems in "dev.acme.com". All of the work systems are linux systems (except my laptop, which is MacOSX).

My user id on the work systems is "astebbens", while my user id on the home systems is "aks".

My work personal laptop is called "somewhere", and my home desktop system is called "anywhere".

Here are the files that can be created to property configure environments appropriate to each combination of user, workgroup, and system.

Filename	Purpose
`.environment.acme.sh`	Define variables for Acme.com
`.enviromment.home.sh`	Define variables for home systems
`.enviromment.prod.acme.com.sh`	Define variables for work production systems
`.enviromment.qa.acme.com.sh`	Define variables for work QA systems
`.enviromment.dev.acme.com.sh`	Define variables for work dev systems
`.enviromment.darwin.sh`	Define variables for Mac OS X systems
`.enviromment.Ubuntu.sh`	Define variables for Ubuntu systems
`.environment.somewhere.sh`	Define variables for work laptop
`.environment.anywhere.sh`	Define variables for home laptop
`.enviromment.astebbens.sh`	Define variables for user astebbens
`.enviromment.aks.sh`	Define variables for user aks

Here is the sequence of files on a production work system:

.environment.sh
.environment.linux.sh
.environment.astebbens.sh
.environment.acme.sh
.environment.prod.acme.com.sh

Here are the files sourced in the work dev environment:

.environment.sh
.environment.linux.sh
.environment.astebbens.sh
.environment.acme.sh
.environment.dev.acme.com.sh

Here are the files sourced in on my work laptop:

.environment.sh
.environment.darwin.sh
.environment.somewhere.sh
.environment.astebbens.sh
.environment.acme.sh
.environment.dev.acme.com.sh

Here are the files sourced in on my home desktop:

.environment.sh
.environment.darwin.sh
.environment.anywhere.sh
.environment.aks.sh
.environment.home.sh

The .environment.sh script does all the work of figuring out which other files need to be sourced.

Environment Variables

The following environment variables are set, possibly based on the given sources.

Envar	Source	Meaning
`USER` `LOGNAME`	`$LOGNAME` `whoami`	User currently logged in
`OSTYPE`	`uname -s`	Operating System type name
`OSREV`	`uname -r`	OS revision
`MACHTYPE`	`uname -m`	Machine type
`HOST`	`uname -n` `hostname`	Host name
`HOSTALASES`	`~/.hostaliases`	Host aliases list
`ORGANIZATION`	`~/.organization`
`WORKGROUP`	`~/.workgroup` `$ORGANIZATION`	Company or organization name
`DOMAIN`	`~/.domain` or end of `uname -n`	The local domain name

DOMAIN and WORKGROUP both come after USER and HOST, so that each of the latter can choose the former, respectively.

Only the files that actually exist are sourced.

After the files are sourced, the file ~/.runenv can be examined to see which files were actually sourced.

Functions Provided

This script provides the following functions to make writing the sub-enviroinment scripts easier:

run_file FILE [RECORDFILE]

source FILE if and only if it exists, if RECORDFILE is given, and if FILE is sourced without errors, then append FILE to the contents of RECORDFILE.

run_env  NAME

If ~/.environment.NAME.sh exists, source it, and if it is sourced without errors, append its name to ~/.runenv

run_env_once NAME

If ~/.environment.NAME.sh exists and has not already been sourced during this session, source it, and record the fact in ~/.runenv.

add_path PATHVAR PATH

Add DIR to the directory list in PATHVAR if it's not already there.

add_path_first PATHVAR DIR

Add (or move) DIR to the front of the PATHVAR.

add_paths PATHVAR options LIST-OF-DIRS ..

Like add_path, all the directories in LIST-OF-DIRS are added to the PATHVAR, using the $SETPATH script, or the add_path function above.

Author

Alan Stebbens aks@stebbens.org

vim: :set ai sw=2 textwidth=66