Re: [PATCH] tests: prefer host Git to verify chainlint self-checks

[Junio C Hamano <gitster@pobox.com>]

Eric Sunshine <sunshine@sunshineco.com> writes:

> (The only postprocessing of "expect" files which needs to stay is the
> bit which removes the "# LINT:" comments which litter the "expect"
> files explaining to human readers why the linter should insert a
> "???FOO???" annotation at that particular point.)

Yup, that matches my understanding.

> The chainlint self-tests were never meant to be about its general
> output stability. They were intended to ensure that the "???FOO???"
> annotations are: (1) indeed inserted for the set of linting problems
> the tool detects, and (2) inserted at the correct spot in the emitted
> output relative to the shell tokens to which the annotation applies.

Yup.

> Minor differences in the tool's output (whether over time or between
> platforms) should be immaterial in respect to those correctness goals.

But there is no reason to make such immaterial changes to the output
gratuitously when we are updating the tool to improve it, no?