From: Junio C Hamano <gitster@pobox.com>
Date: Wed, 25 Nov 2020 20:49:47 GMT
Subject: Re: [PATCH v4] help.c: expand options for help.autocorrect
Message-ID: <xmqqk0u9duvo.fsf@gitster.c.googlers.com>
In-Reply-To: <20201125183309.GF8396@szeder.dev>

SZEDER Gábor <szeder.dev@gmail.com> writes:

>> +		grep "is not a git command" actual &&
>> +		test_line_count = 1 actual
>
> The two branches of this condition are in the wrong order, as it
> doesn't check the expected error message in normal mode, but it does
> grep for the specific translated error message when run with
> GIT_TEST_GETTEXT_POISON=1 which then fails the test.

Correct.

> I think we should simply use 'test_i18ngrep' instead of that whole
> if-else block.  
> ...
> I wonder what that 'test_line_count = 1' is supposed to check, and am
> not sure that it's really necessary.

I think it is to make sure that there is no other message emitted.

I prefer not to rely on the current implementation of poison that
makes any message a single line message "# GETTEXT POISON #" that is
an incomplete line.  In other words,

	main() {

	printf(_("%s.\n"), "hello");
	error(_("goodbye"));

	exit(0);
	}

would give you a single line output with the current implementation,
but there is no guarantee we won't improve the poisoning in the
future.