notebook/notes/bash/robustness.md

title	TARGET DECK	FILE TAGS	tags
Measuring Command Robustness	Obsidian::STEM	bash	bash shell

Overview

An interesting point Robbins discusses in his introduction to posix/awk/index is this idea of command robustness. He states that:

A self-contained shell script is more reliable because there are no other files to misplace.

%%ANKI Basic What is a self-contained shell script? Back: A shell script that does not rely on the presence of other files. Reference: Robbins, Arnold D. “GAWK: Effective AWK Programming,” October 2023. https://www.gnu.org/software/gawk/manual/gawk.pdf

END%%

%%ANKI Basic What makes a self-contained shell script more reliable? Back: There are no other files to misplace. Reference: Robbins, Arnold D. “GAWK: Effective AWK Programming,” October 2023. https://www.gnu.org/software/gawk/manual/gawk.pdf

END%%

He argues that the first command below is more robust than the second since the command is more loosely coupled to its environment:

$ awk 'program' input-file1 input-file2 ...
$ awk -f program-file input-file1 input-file2 ...

It's interesting to think what else can be used as a measure of a command's robustness:

• Required environment variables or environment variables formatted in a certain way
  • $PATH needs to point to a specific location
• Whether the invoked program is present by default in a distribution or must be installed
• The ability to run with or without an associated TTY
• The ability to run with or without associated standard streams
  • stdout, stderr, stdin
• How backwards compatible the invoked program is
  • Version mismatch may silently cause the same invocation to fail
• Expected permissions
  • EUID, read permissions on an input file, etc.
• Determinism of the program itself
  • Does output rely entirely on input or can it make nondeterministic choices
• The amount of resources dedicated to the program
  • Failure may occur if not enough memory is provided to the command
• Whether a program acts idempotently
  • What happens if I run the command twice in a row?
• Whether a program acts atomically
  • Is it possible intermediate files are left that affect subsequent runs?
• The presence of timeouts
  • Perhaps a program waits a specified amount of time before input is available. The command's success is now externally determined.
• Locale-aware functionality
  • Consider for instance posix/awk/index's \u escape-sequences which targets characters in the current locale's character set as opposed to specifically Unicode.
  • POSIX standard posix/regexp#Character Classes serve as another example.

The above scenarios are what makes something like nix/index so compelling.

Bibliography

• Robbins, Arnold D. “GAWK: Effective AWK Programming,” October 2023. https://www.gnu.org/software/gawk/manual/gawk.pdf