I would just use sh syntax and source it.

Config file:

Code:

$ cat config
PATHS_TO_WATCH='/home/ /var/www/'
LIMITED_TRANSFER_RATE=800

Read config:

Code:

$ . ./config
$ echo "$PATHS_TO_WATCH"
/home/ /var/www/
$ echo "$LIMITED_TRANSFER_RATE"
800

If you want the script to abort if there are any invalid variable names in the config (if you try the following, with an invalid variable name in config, in your current shell, it will *poof* disappear , so make sure to do it in a subshell):

Code:

set -e
. config_file
set +e

Or, you could leave "set -e" on, if you like (it causes the shell to terminate when a command fails, with certain exceptions regarding and-or lists and conditionals ... read your shell's 'set' documentation for more info).

As to spaces in a pathname, I don't see how you can detect them since you are using spaces to delimit each path. How is the code to know if the space is embedded in a path or meant to separate one path from another? To be absolutely certain of whether a character is in a path or a delimiter, you would need to select a delimiter that cannot occur in a UNIX path. There is only one such character, the null byte, and it's notoriously difficult to work with in most shells. Perhaps simply checking the validity of each path is sufficient.

Assuming that the config file was sourced as in my example above (using the dot command):

Code:

for p in $PATHS_TO_WATCH; do
    if [ ! -d $p ]; then
        echo $0: $p: bad path 1>&2
        exit 1
    fi
done

If an invalid path is encountered, a message (including the script's name and the offending path) is printed to standard error and an exit status of 1 is returned.

Cheers,
Alister