Path: csiph.com!feeder.erje.net!2.eu.feeder.erje.net!ecngs!feeder2.ecngs.de!78.46.240.70.MISMATCH!weretis.net!feeder4.news.weretis.net!news.mixmin.net!aioe.org!bofh.it!news.nic.it!robomod From: Samuel Thibault Newsgroups: linux.debian.maint.python Subject: Re: Best way to handle circular build deps to make a pypy- package Date: Wed, 15 Aug 2018 17:40:01 +0200 Message-ID: References: X-Mailbox-Line: From debian-python-request@lists.debian.org Wed Aug 15 15:30:44 2018 Old-Return-Path: X-Amavis-Spam-Status: No, score=-6.899 tagged_above=-10000 required=5.3 tests=[BAYES_00=-2, FOURLA=0.1, HEADER_FROM_DIFFERENT_DOMAINS=0.001, LDO_WHITELIST=-5] autolearn=ham autolearn_force=no X-Policyd-Weight: using cached result; rate: -6.1 Mail-Followup-To: Samuel Thibault , Pierre-Elliott =?utf-8?B?QsOpY3Vl?= , debian-python@lists.debian.org MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit Organization: I am not organized User-Agent: NeoMutt/20170113 (1.7.2) X-Mailing-List: archive/latest/16228 List-ID: List-URL: List-Archive: https://lists.debian.org/msgid-search/20180815152944.fqxdvnahfpx3vkbi@var.youpi.perso.aquilenet.fr Approved: robomod@news.nic.it Lines: 32 Sender: robomod@news.nic.it X-Original-Cc: debian-python@lists.debian.org X-Original-Date: Wed, 15 Aug 2018 17:29:44 +0200 X-Original-Message-ID: <20180815152944.fqxdvnahfpx3vkbi@var.youpi.perso.aquilenet.fr> X-Original-References: <20180815110509.nyvuqbncnlm54pp7@crans.org> <20180815111039.xl62gs2cibvqoolm@var.youpi.perso.aquilenet.fr> <20180815131125.sx5fcgnapb4wuhow@crans.org> <20180815134318.j6iq3i3hntnygrpm@crans.org> Xref: csiph.com linux.debian.maint.python:10808 Hello, Pierre-Elliott Bécue, le mer. 15 août 2018 15:47:36 +0200, a ecrit: > So I found a way to build locally, using your suggestions. In my case I > guess I could upload directly the output of the nocheck build as the bin > package isn't altered. Yes, but better request a binnmu after the upload, to be sure. > But how should I handle the upload in the archive in a general case? Let's > imagine my profile produces a different bin package, eg if foo -> bar (for > specific foo functionalities) -> foo, so with a stage1 foo, then a stage1 > bar and then a stage2 foo, how should I proceed? Do all the builds locally > and upload the packages obtained from the last build? AIUI it is usually preferred to upload as few manually-built packages as possible, and thus you'd upload stage1 foo, let bar be built, and request a binmu for stage2 foo. > On the other hand, let's assume one wants to do a sourceful upload, and not > upload the .deb files (I know this is not possible when one introduces a new > package in the archive), then the buildd farm couldn't succeed because the > nocheck wouldn't be taken into account. So this means profile builds > systematically require source + binary upload? If the profile-specific build really is needed, yes. In practice, once the loop is broken it's not needed since you already have a stage1 bar to build with and so you can just let builds happen. It's then only for new archs that a bootstrap is needed (and keeping the profile available for doing it is really useful for porters to just do it themselves). Samuel