[Bug / regression?] In file ignore directives (# shellcheck disable=<check_id>) are not specific & disable all checks instead (not just check_id)

[fedy-cz]

Shellcheck version: 0.9.0

Stumbled upon this randomly, but it seems that shellcheck (v 0.9.0) ignores the rule numbers/IDs and just disables all the checks when ignore directives are used inside a file.

For example, here I try to ignore the rule SC2251:

#!/bin/bash
set -o errexit -o pipefail -o noclobber -o nounset

# -allow a command to fail with !’s side effect on errexit
# -use return value from ${PIPESTATUS[0]}, because ! hosed $?
#shellcheck disable=SC2251
! getopt --test > /dev/null
if [[ ${PIPESTATUS[0]} -ne 4 ]]; then
  echo "I’m sorry, 'getopt --test' failed in this environment." >&2
  exit 1
fi

... So far so good.

But if I change the line from:
#shellcheck disable=SC2251
to
#shellcheck disable=SC1000

I still get no errors.

Other versions affected:
https://www.shellcheck.net/ seems to be affected as well.

[koalaman]

Nice catch! This was due to one of ShellCheck's design mistakes: making directives separate nodes in the AST that checks match against to identify problems.

In this case, the check looked for ! cmd inside various compound command bodies. It neglected to look for ! cmd inside directives inside the compound command bodies. As a result, any type of directive (whether or not related to disabling warnings) would cause this particular check not to recognize the problem it aimed to.

It was fortunately not a general problem with disabling specific codes, but it's definitely not the first bug caused by this design mistake, and it certainly won't be the last.