Re: [PATCH 1/3] t3301: tolerate minor notes-related presentation changes

From: Eric Sunshine <sunshine@sunshineco.com>
Date: Aug 30, 2021, 18:16 UTC
Message-ID: <CAPig+cRqMjaujwfwzrqNyxaMN9y4r7RP2W6kOcP-kKcnvrjYxQ@mail.gmail.com>
In-Reply-To: <xmqq4kb66er7.fsf@gitster.g>

On Mon, Aug 30, 2021 at 1:09 PM Junio C Hamano <gitster@pobox.com> wrote:

Show 11 quoted lines

> Eric Sunshine <sunshine@sunshineco.com> writes:
> > These tests care about whether intended notes-related functionality
> > occurred and that `git log` presents the notes in the expected fashion
> > (or, in some cases, that `git log` suppresses the notes). However, the
> > tests hard-code the precise indentation of notes by the default `git
> > log` output, which makes them somewhat brittle since they won't be able
> > to tolerate even minor changes to the presentation. Make the tests a bit
> > more robust by ignoring indentation.
>
> Isn't this losing too much information?  If we lose all, or gain
> random number of, leading whitespaces, the test won't notice.

That was the idea. The precise amount of indentation -- whether four spaces or one TAB or whatever -- seems pretty much immaterial in the wide view [1], and it is not inconceivable that the exact amount of indentation might change in the future, thus this future-proofs the tests against minor indentation changes.

However, as mentioned in my response to Ævar, I wavered quite a bit on whether or not to make this change since, although the justification of "future-proofing" the tests isn't exactly hand-wavy, we don't need the change either. It may be a case of YAGNI. So, as I also mentioned in that response, I don't mind at all dropping this patch and going with Ævar's version.

[1]: mechanical extraction aside...