From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:52394) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1beQiE-0006fa-9n for qemu-devel@nongnu.org; Mon, 29 Aug 2016 13:51:11 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1beQi9-0008Gi-Ap for qemu-devel@nongnu.org; Mon, 29 Aug 2016 13:51:09 -0400 Received: from mail.weilnetz.de ([37.221.199.173]:56298) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1beQi9-0008GZ-4j for qemu-devel@nongnu.org; Mon, 29 Aug 2016 13:51:05 -0400 References: <1471975876-23385-1-git-send-email-sw@weilnetz.de> <87bn0jjoos.fsf@oc4731375738.ibm.com> <871t1fjlv1.fsf@oc4731375738.ibm.com> <87k2ezioj7.fsf@oc4731375738.ibm.com> From: Stefan Weil Message-ID: <92b7dd7e-ecd7-6c1d-95b9-49d62c9aa45f@weilnetz.de> Date: Mon, 29 Aug 2016 19:51:01 +0200 MIME-Version: 1.0 In-Reply-To: <87k2ezioj7.fsf@oc4731375738.ibm.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH for-2.7] wxx: Fix broken build (mkdtemp unavailable) List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Sascha Silbe , Peter Maydell Cc: QEMU Developer On 08/29/16 17:27, Sascha Silbe wrote: > Dear Peter, > > Peter Maydell writes: > >> On 25 August 2016 at 10:36, Peter Maydell w= rote: >>> On 23 August 2016 at 16:01, Sascha Silbe w= rote: >>>> Glad to hear. It would be possibly to support the combination of gli= b < >>>> 2.30.0 AND windows, but only by copying a considerable amount of cod= e >>>> from glib. I'd prefer to avoid that if we can help it. >>> >>> If we want to raise the minimum glib version requirement for >>> Windows we need to enforce this in configure. (We have had >>> a higher minimum for Windows hosts in the past, so there's >>> precedent for doing it.) >> >> Or we could arrange to skip this test if we're on windows >> with an old glib I guess. > > In general I agree with you. In practice test-logging was completely > broken on Windows since 2.6.0 (it hard-coded /tmp) and I don't have a > suitable environment to test a Windows build, so I'd feel uncomfortable > submitting patches addressing this issue myself. > > However that shouldn't stop anyone else (Stefan perhaps? :) ) from > fixing the tests on Windows. I'll gladly review the effects of the > corresponding patches on the POSIX side. > > Sascha For 2.7, I think we don't need this patch nor any other solution,=20 because the problem is less critical than I thought initially: only=20 builds with old versions of glib are affected. For 2.8, raising the minimum glib version to 2.30.0 would avoid the=20 build problem for normal builds targeting Windows. IMHO it is the=20 simplest and also an acceptable solution, and it also allows removing=20 some conditional code. Fixing the tests for builds targeting Windows is a different issue.=20 Ideally "make test" should work for such builds, no matter whether they=20 run as cross build on Linux (my usual environment) or native on Windows.=20 To simplify things, some tests which are difficult to fix for Windows=20 and which don't test Windows specific code could be omitted. I cannot=20 promise that I'll work on test support for Windows in the near future =E2= =80=93=20 if anybody else does that job, I'll be happy, too. Stefan