Path: csiph.com!newsfeed.xs4all.nl!newsfeed7.news.xs4all.nl!bofh.it!news.nic.it!robomod From: Stuart Prescott Newsgroups: linux.debian.maint.python Subject: Re: .egg-info for entry points during dh_auto_test Date: Fri, 22 Oct 2021 09:20:01 +0200 Message-ID: References: X-Mailbox-Line: From debian-python-request@lists.debian.org Fri Oct 22 07:12:00 2021 Old-Return-Path: X-Amavis-Spam-Status: No, score=-6.751 tagged_above=-10000 required=5.3 tests=[BAYES_00=-2, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HEADER_FROM_DIFFERENT_DOMAINS=0.249, LDO_WHITELIST=-5, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham autolearn_force=no X-Policyd-Weight: NOT_IN_SBL_XBL_SPAMHAUS=-1.5 CL_IP_EQ_HELO_IP=-2 (check from: .nanonanonano. - helo: .mail-pg1-x52c.google. - helo-domain: .google.) FROM/MX_MATCHES_NOT_HELO(DOMAIN)=0; rate: -3.5 X-Gm-Message-State: AOAM533+d68RriDZyKuq9rnTyfPOeO0JYhGpjI5bEccKqtuskGQqGhd1 JkGyM7U2VCy+mrEIr7hvzAcmFGQL3OeOSw== X-Google-SMTP-Source: ABdhPJwr88hEBqXTsEUDKV9ZMEU/iSCSzGZifOIKPCo2Px2FiD2/wO7PPt86rpV/q6KUCTznMJ8ZEw== X-Received: by 2002:a65:62c1:: with SMTP id m1mr8240401pgv.339.1634886697920; Fri, 22 Oct 2021 00:11:37 -0700 (PDT) Sender: robomod@news.nic.it MIME-Version: 1.0 Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="utf-8" X-Mailing-List: archive/latest/19445 List-ID: List-URL: List-Archive: https://lists.debian.org/msgid-search/2908703.xIg9TgxuJX@simurgh Approved: robomod@news.nic.it Lines: 36 Organization: linux.* mail to news gateway X-Original-Cc: debian-python@lists.debian.org X-Original-Date: Fri, 22 Oct 2021 18:11:32 +1100 X-Original-Message-ID: <2908703.xIg9TgxuJX@simurgh> X-Original-References: <33f3616e-50d7-93d4-4b21-ae818a6dec6f@fladi.at> <20211022034414.2bmtenuabf2qv2hw@satie.tumbleweed.org.za> X-Original-Sender: Stuart Prescott Xref: csiph.com linux.debian.maint.python:13649 Hi Michael I wrestled with something similar some years ago with translate-toolkit, which has a set of tests that run the binaries it ships but these (entry-points) binaries are not generated until after installation. The pybuild/debhelper sequence is to build, test then install; upstream's view is to build, install then test, and so that is what I faked for that package, overriding the normal test invocation, and running the tests in an execute_after_dh_install step: https://sources.debian.org/src/translate-toolkit/unstable/debian/rules/#L59 I help the test suite find the binaries in debian/translate-toolkit/usr/bin and pybuild points the test harness at the modules in the normal place in .pybuild/. That's pretty ugly, but it has worked out ok for this package. A related problem is the PEP517 builders go from source to wheel and then install the wheel; the entry_points and scripts only appear after installation, and so once again, testing most probably lives after 'install' not before. Given that entry_points, plugin registration, PEP517 are all pointing us towards build/install/test as the paradigm rather than build/test/install, do we need to look for an additional dh step to be added to our normal sequence to help us with what is becoming a common task? regards Stuart -- Stuart Prescott http://www.nanonanonano.net/ stuart@nanonanonano.net Debian Developer http://www.debian.org/ stuart@debian.org GPG fingerprint 90E2 D2C1 AD14 6A1B 7EBB 891D BBC1 7EBB 1396 F2F7