Re: [PATCH v2 1/5] split-index: add tests to demonstrate the racy split index problem

From: Junio C Hamano <gitster@pobox.com>
Date: Sep 28, 2018, 17:30 UTC
Message-ID: <xmqqo9chcyxi.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <20180928004843.GP27036@localhost>

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

Show 25 quoted lines

> Junio,
>
> On Thu, Sep 27, 2018 at 02:44:30PM +0200, SZEDER Gábor wrote:
>> diff --git a/t/t1701-racy-split-index.sh b/t/t1701-racy-split-index.sh
>> new file mode 100755
>> index 0000000000..ebde418d7e
>> --- /dev/null
>> +++ b/t/t1701-racy-split-index.sh
>> @@ -0,0 +1,218 @@
>> +#!/bin/sh
>> +
>> +# This test can give false success if your machine is sufficiently
>> +# slow or all trials happened to happen on second boundaries.
>> +
>> +test_description='racy split index'
>> +
>> +. ./test-lib.sh
>> +
>> +test_expect_success 'setup' '
>> +	# Only split the index when the test explicitly says so.
>> +	sane_unset GIT_TEST_SPLIT_INDEX GIT_FSMONITOR_TEST &&
>
> Please note that this patch adds another use of the environment
> variable GIT_FSMONITOR_TEST, while the topic 'bp/rename-test-env-var'
> is about to rename that variable to GIT_TEST_FSMONITOR.

I think the most sensible thing to do during transition is to set (or unset) both consistently (and leave a note in t/test-lib.sh near check_var_migration that t1701 has done so and the dup should be corrected when transition is over---but that is optional).