Bash functions return an integer value. If you want to return something else then you'll have to use a global variable. And that means documenting the return variable and choosing something unlikely to conflict with a variable that the author of any consuming script might use for another purpose. Tricky and fragile.
It turns out that bash has a separate namespace for functions and variables (thanks to http://stackoverflow.com/a/30042245/1392864). So, if the function just sets a global variable having the same name as the function, then we have a nice self-documenting naming convention for function return variables. And it is also less likely that there will be a conflict between script variables and the function return variable names (because functions tend to have verb or verb-noun names, while variables tend to have noun names).
An example:
parse_version() {
# Initialize result and local variables
#
local STD_VSN_=( "0" "0" "0" "" )
parse_version=( ${STD_VSN_[@]} )
# ... do stuff ...
# Set global $parse_version to result and
# get out
#
parse_version=( ${STD_VSN_[@]} )
return 0
}
parse_version "1.2.3-p456"
if (( 0 != $? )); then
echo "ERROR: failed to parse version."
exit 2
fi
VSN_PARTS_=( ${parse_version[@]} )
echo "MAJOR: ${VSN_PARTS_[0]}"
echo "MINOR: ${VSN_PARTS_[1]}"
echo "PATCH: ${VSN_PARTS_[2]}"
echo " SUB: ${VSN_PARTS_[3]}"© 2016 Dave Hein
This work by Dave Hein is licensed under a Creative Commons Attribution 4.0 International License.
