Re: [PATCH v2 3/3] t/lib-httpd: avoid occasional failures when checking access.log

From: Jeff King <peff@peff.net>
Date: Jul 11, 2018, 14:57 UTC
Message-ID: <20180711145730.GB15269@sigill.intra.peff.net>
In-Reply-To: <20180711125647.16512-4-szeder.dev@gmail.com>

On Wed, Jul 11, 2018 at 02:56:47PM +0200, SZEDER Gábor wrote:

Show 11 quoted lines

> +# Requires one argument: the name of a file containing the expected stripped
> +# access log entries.
> +check_access_log() {
> +	sort "$1" >"$1".sorted &&
> +	strip_access_log >access.log.stripped &&
> +	sort access.log.stripped >access.log.sorted &&
> +	if ! test_cmp "$1".sorted access.log.sorted
> +	then
> +		test_cmp "$1" access.log.stripped
> +	fi
> +}

This will generate output showing both the unsorted and sorted differences. Do we want to suppress the sorted ones (e.g., by just using "cmp" directly)? I guess it doesn't matter unless there is an actual test failure, but I just wonder if it would be confusing to see both.

Show 22 quoted lines

> diff --git a/t/t5541-http-push-smart.sh b/t/t5541-http-push-smart.sh
> index 6cd986797d..a481e3989a 100755
> --- a/t/t5541-http-push-smart.sh
> +++ b/t/t5541-http-push-smart.sh
> @@ -43,15 +43,13 @@ test_expect_success 'no empty path components' '
>  	cd "$ROOT_PATH" &&
>  	git clone $HTTPD_URL/smart/test_repo.git/ test_repo_clone &&
>  
> -	strip_access_log >act &&
> +	check_access_log exp
> +'
>  
> +test_expect_success 'clear access log' '
>  	# Clear the log, so that it does not affect the "used receive-pack
>  	# service" test which reads the log too.
> -	#
> -	# We do this before the actual comparison to ensure the log is cleared.
> -	>"$HTTPD_ROOT_PATH"/access.log &&
> -
> -	test_cmp exp act
> +	>"$HTTPD_ROOT_PATH"/access.log
>  '

This took some head-scratching, mostly because of the original comment. I thought the "before" here was referring to a timing issue (maybe because we're talking about timing ;) ).

But it is really "make sure that a failed test here does not prevent us from doing this cleanup". So the original really should have just dropped that comment and added a test_when_finished. Bumping it into a separate test as you have here accomplishes the same thing.

-Peff