From: Junio C Hamano <gitster@pobox.com>
Date: Wed, 21 Sep 2011 02:34:29 GMT
Subject: Re: [PATCH 1/5] git-p4 tests: refactor, split out common functions
Message-ID: <7vfwjq4o96.fsf@alter.siamese.dyndns.org>
In-Reply-To: <20110921012917.GA27256@arf.padd.com>

Pete Wyckoff <pw@padd.com> writes:

> Yes, this is all very shoddy, and I suspect that I don't need the
> initial kill_p4d as long as the other tests fail nicely.
>
> But your point about killing the _wrong_ p4d is good.  But p4d
> isn't friendly like other daemons that will drop a pid file for
> help in future cleanup.  I can run it without the "-d", in which
> case it does not daemonize.  Then I should be able to use shell
> job control to kill off the one I started, hopefully.

Yeah, something like

	: start
	p4d &
        echo $! >$pidfile

        : stop
	while kill -0 $pidfile
        do
	        kill $pidfile
                sleep 1
	done

> Finding an unused port is also a pain.

I think the way t55xx series assign LIB_HTTPD_PORT is an acceptable
compromise.