From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [KVM-AUTOTEST PATCH] KVM test: refactor kvm_config.py Date: Wed, 09 Feb 2011 12:19:23 +0200 Message-ID: <4D526A2B.5050303@redhat.com> References: <1297216207-28375-1-git-send-email-mgoldish@redhat.com> <4D525E58.4080009@redhat.com> <4D526755.8080903@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: autotest@test.kernel.org, kvm@vger.kernel.org, Uri Lublin To: Michael Goldish Return-path: Received: from mx1.redhat.com ([209.132.183.28]:59775 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753392Ab1BIKT2 (ORCPT ); Wed, 9 Feb 2011 05:19:28 -0500 In-Reply-To: <4D526755.8080903@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: On 02/09/2011 12:07 PM, Michael Goldish wrote: > On 02/09/2011 11:28 AM, Avi Kivity wrote: > > On 02/09/2011 03:50 AM, Michael Goldish wrote: > >> This is a reimplementation of the dict generator. It is much faster > >> than the > >> current implementation and uses a very small amount of memory. > >> Running time > >> and memory usage scale polynomially with the number of defined variants, > >> compared to exponentially in the current implementation. > >> > >> Instead of regular expressions in the filters, the following syntax is > >> used: > >> > >> , means OR > >> .. means AND > >> . means IMMEDIATELY-FOLLOWED-BY > >> > >> Example: > >> > >> only qcow2..Fedora.14, RHEL.6..raw..boot, smp2..qcow2..migrate..ide > >> > > > > > > Is it not possible to keep the old syntax? Breaking people's scripts is > > bad. > > No, because the old syntax uses regexps and there's no clean way to > prune tree branches early if those are supported. > Ok. > For users who have their own tests_base.cfg (if there are any), we may > have to keep the old parser as an alternative, or detect the presence of > an incompatible cfg file and warn about it. Does that sound like a good > idea? No. It increases the maintenance burden and user confusion. -- error compiling committee.c: too many arguments to function