From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Schulte Subject: Re: FYI: Org mode testing framework, Emacs 23 and 22 Date: Wed, 09 Nov 2011 14:22:04 -0700 Message-ID: <8739dxdm9v.fsf@gmail.com> References: <874o0veejl.wl%dmaus@ictsoc.de> <87y5y7oxyt.fsf@gmail.com> <877h5f558c.wl%dmaus@ictsoc.de> <87aaaaokpq.fsf@gmail.com> <87botakwru.fsf@gnu.org> <80hb32w4sp.fsf@somewhere.org> <87vcrhyp5g.fsf@gmail.com> <801uu2oagc.fsf@somewhere.org> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Return-path: Received: from eggs.gnu.org ([140.186.70.92]:56845) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ROFb4-0000gK-Ig for emacs-orgmode@gnu.org; Wed, 09 Nov 2011 16:22:15 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ROFb2-0001nb-Si for emacs-orgmode@gnu.org; Wed, 09 Nov 2011 16:22:14 -0500 Received: from mail-qw0-f41.google.com ([209.85.216.41]:55982) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ROFb2-0001nW-J1 for emacs-orgmode@gnu.org; Wed, 09 Nov 2011 16:22:12 -0500 Received: by qadc11 with SMTP id c11so2265620qad.0 for ; Wed, 09 Nov 2011 13:22:12 -0800 (PST) In-Reply-To: (Martyn Jago's message of "Tue, 25 Oct 2011 18:40:50 +0100") List-Id: "General discussions about Org-mode." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: emacs-orgmode-bounces+geo-emacs-orgmode=m.gmane.org@gnu.org Sender: emacs-orgmode-bounces+geo-emacs-orgmode=m.gmane.org@gnu.org To: Martyn Jago Cc: emacs-orgmode@gnu.org Martyn Jago writes: > Hi > > Brian Wightman writes: > >> On Mon, Oct 24, 2011 at 7:12 AM, Sebastien Vauban >> wrote: >>> For my information, why do you need to test that 2 suites don't run at = the >>> same time? =C2=A0They only write to temp buffers, no? =C2=A0Can they co= nflict? >> >> If they do conflict and only one set of tests should run at a time, >> shouldn't the second set of tests be queued up so it runs when the >> first is complete? If this isn't done, I could see a situation where >> at least one commit remains untested until the next commit. >> >> my $0.02; >> Brian > > I would think it would make sense to ensure the newly committed code > also builds correctly (including info), and why not test back to Emacs > 22 also (since that is how this thread came about)? I have set up > something similar, and doing all this takes less than 1 minute on my > machine. > > Incidentally, I added info on running tests on Emacs 22 and 23 to Worg at= =20 > > http://orgmode.org/worg/org-tests/index.html > These are good points, so the best solution would have two parts, the first should en-queue each new commit as it is received, and the second should pop commits off of the queue, run the test suite on that commit and write the results to an html file. My first thought for a shell-script solution would use a mkfifo named pipe, something like the following... ,----[setup] | mkfifo /tmp/org-commits-to-test `---- ,----[in a post-commit hook] | echo $COMMIT >> /tmp/org-commits-to-test `---- ,----[in another constantly-running shell script] | while true; do | for i in $(cat /tmp/org-commits-to-test);do | cd /testing/repo/location | git checkout $i | git reset --hard HEAD | emacs --batch ... | done | done `---- But I imagine there is likely a better solution. Best -- Eric --=20 Eric Schulte http://cs.unm.edu/~eschulte/