A bash scripting template incorporating best practices & several useful functions.
I write bash scripts not infrequently and realised that I often copied a recent script whenever I started writing a new one. This provided me with a basic scaffold to work on and several useful helper functions I'd otherwise likely end up duplicating.
So rather than continually copying old scripts and flensing the irrelevant code, I'm publishing a more formalised template to ease the process for my own usage and anyone else who may find it helpful! Suggestions for improvements are most welcome.
| File | Description |
|---|---|
| template.sh | A fully self-contained script which combines source.sh & script.sh. |
| source.sh | Designed for sourcing into scripts; contains only those functions unlikely to need modification. |
| script.sh | Sample script which sources in source.sh and contains those functions likely to be modified. |
| build.sh | Generates template.sh by combining source.sh & template.sh. Just a helper script for me. |
The Bash scripting community is an opinionated one. This is not a bad thing, but it does mean that some decisions made in this template aren't going to be agreed upon by everyone. A few of the most notable ones are highlighted here with an explanation of the rationale.
Conventional wisdom has for a long time held that at the top of every Bash script should be set -e (or the equivalent set -o errexit). This modifies the behaviour of Bash to exit immediately when it encounters a non-zero exit code from an executed command if it meets certain criteria. This would seem like an obviously useful behaviour in many cases, however, controversy arises both from the complexity of the grammar which determines if a command is eligible for this behaviour and the fact that there are many circumstances where a non-zero exit code is expected and should not result in termination of the script. An excellent overview of the argument against this option can be found in BashFAQ 105.
My personal view is that the benefits of errexit outweigh its disadvantages. More importantly, a script which is compatible with this option will work just as well if it is disabled, however, the inverse is not true. By being compatible with errexit those who find it useful can use this template without modification while those opposed can simply disable it without issue.
By enabling set -u (or the equivalent set -o nounset) the script will exit if an attempt is made to expand an unset variable. This can be useful both for detecting typos as well as potentially premature usage of variables which were expected to have been set earlier. The controvery here arises in that many Bash scripting coding idioms rely on referencing unset variables, which in the absence of this option are perfectly valid. Further discussion on this option can be found in BashFAQ 112.
This option is enabled for the same reasons as described above for errexit.
All content is licensed under the terms of The MIT License.