Re: [PATCH 1/9] git p4 test: wait longer for p4d to start

From: Junio C Hamano <gitster@pobox.com>
Date: Jun 26, 2012, 03:46 UTC
Message-ID: <7vbok6enw3.fsf@alter.siamese.dyndns.org>
In-Reply-To: <1340673505-10551-2-git-send-email-pw@padd.com>

Pete Wyckoff <pw@padd.com> writes:

> Running tests at high parallelism on a slow machine, 5 sec is
> not enough to wait for p4d to start.  Change it to 10 sec.
>
> Signed-off-by: Pete Wyckoff <pw@padd.com>

I'd rather not to see this kind of incremental micro-tweaks. The next person who runs on an even slower box or with more excessive parallelism compared to his machine size may complain and send a patch to raise this to 20.

Who are we trying to help, and if you raised this to "forever", whom would such a change be hurting? If there is something wrong with the p4 installation, it may never come up, but an interactive user will eventually end up seeing "waiting for p4d to start" and nothing else in such a case, and will know to kill it with ^C. An automated test spawned from a cron job would definitely be hurt with a timeout of "forever", but to them, 10 seconds or 10 minutes would not make much of practical differences (as long as you do not run such a cron job too often).

My preference would probably be a very long default that can be shortened (or lengthened) with an environment variable, so that a frequent auto-builder that runs the tests every five minutes can tune it down to 30 seconds, or something like that.

Show 16 quoted lines

>  t/lib-git-p4.sh | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/t/lib-git-p4.sh b/t/lib-git-p4.sh
> index 121e380..0080eae 100644
> --- a/t/lib-git-p4.sh
> +++ b/t/lib-git-p4.sh
> @@ -37,7 +37,7 @@ start_p4d() {
>  		p4d -q -r "$db" -p $P4DPORT &
>  		echo $! >"$pidfile"
>  	) &&
> -	for i in 1 2 3 4 5 ; do
> +	for i in 1 2 3 4 5 6 7 8 9 10 ; do
>  		p4 info >/dev/null 2>&1 && break || true &&
>  		echo waiting for p4d to start &&
>  		sleep 1