Re: [PATCH 3/4] t7401: ensure uniformity in the '--for-status' test

From: Junio C Hamano <gitster@pobox.com>
Date: Aug 6, 2020, 17:18 UTC
Message-ID: <xmqqimdvlml9.fsf@gitster.c.googlers.com>
In-Reply-To: <20200806085043.GB13942@konoha>

Shourya Shukla <shouryashukla.oo@gmail.com> writes:

Show 18 quoted lines

> On 05/08 02:30, Junio C Hamano wrote:
>> Shourya Shukla <shouryashukla.oo@gmail.com> writes:
>> 
>> > The '--for-status' test got its expected output from stdin. This is
>> > inconsistent with the other tests in the test script which get their
>> > expected output from a file named 'expected'.
>> >
>> > So, change the syntax of the '--for-status' test for uniformity.
>> 
>> There are a handful examples in t5401 and another one in t3700 that
>> give the "golden master" from the standard input.  When the expected
>> output is used only once, I do not think it is particularlly bad to
>> have it this way.  So,... meh?
>
> I realised what you were trying to say after checking out t5400 and
> t3700. I understand that this change may not be immediately needed but I
> think it does make reading the diff a bit easier since having a '-' as a
> file name does get a bit confusing when reading the output.

If so, perhaps justifying the change based on that, not on "consistency", would be a good idea.

        Side note: But would that mean you'd find it "confusing" to
        read output from 3700 and 5400?  Would "test writers should
        get used to it" be a workable alternative solution?

Since "test_cmp expect actual" and "test_cmp - actual <<HERE" are _both_ valid forms that are useful for different situations, I do not see a compelling reason to insist on one form is consistently used and ban the use of the other.

So...