ostree: Drop

[martinpitt]

This moved into a separate project:
https://github.com/cockpit-project/cockpit-ostree

---

As the standalone cockpit-ostree 175-1 is in Fedora rawhide now, the next release of Cockpit should not ship that any more, to avoid version conflicts and duplicate builds. I will also upload the standalone one to Fedora 28 as soon as the dist-git branch is created.

• create f28 Fedora dist-git branch (https://pagure.io/releng/fedora-scm-requests/issue/7563)
• upload cockpit-ostree to Fedora 28
• fix semaphore race condition in pyflakes (PR tools: Ignore dangling test/avocado/testvm.py symlink in pyflakes check #9789)

[martinpitt]

Priority as this at least needs to be considered for the next release, depending on the fedora 28 progress.

[martinpitt]

fedora 28 dist-git branch now exists, package built and submitted to bodhi: https://bodhi.fedoraproject.org/updates/FEDORA-2018-0effffd2d7

[martinpitt]

Semapore fails with

FAIL: tools/test-static-code 2 pyflakes test

I cannot reproduce that locally, there it works for me. But I think this is during distcheck, so apparently still something wonky there.

Update: make distcheck works for me locally, so need to test this in a cockpit/tests container.

[martinpitt]

I cannot reproduce this in a container either, also not locally with a separate build tree:

$ mkdir build; cd build
$ ../autogen.sh --prefix=/usr --sysconfdir=/etc --localstatedir=/var --enable-debug
$ srcdir=.. ../tools/test-static-code

This works fine. But I still have trouble to see in the log what actually failed, as there's no actual output from pyflakes, and this PR doesn't touch this part at all. The only error message that I see is

test/avocado/testvm.py: No such file or directory

and indeed I don't see that in other semaphore runs. This smells like a weird race condition which has been there for a long time?

[martinpitt]

I have an idea what's going on. test/avocado/testvm.py is a dangling symlink in the release tarballs, as they point to ../bots/ which we don't ship (by intent). So the question is rather, how can test-static-code and pyflakes sometimes do work in distcheck -- either it should always fail on the dangling symlink, or never.

[martinpitt]

Oooooh, I know! Running pyflakes without the grep -Ev filter shows:

$ pyflakes3 test/ test/verify/check-* 2>&1|grep testvm
test/verify/check-storage-used:23: 'from storagelib import *' used; unable to detect undefined namestest/avocado/testvm.py: No such file or directory

So that error message just sometimes gets hidden because it interleaves with the pyflakes errors. But sometimes, if it is on a different line, we catch it.

[martinpitt]

I sent PR #9789 to fix the flaky semaphore test, marking blocked as this needs to land after that to quiesce semaphore.

[croissanne]

This looks good (and satisfying).

One question though: I grepped through the commit and there's a bunch of references to the ostree package in the po files. Will these get removed or are the po files just identical across all cockpit-independent?

[martinpitt]

@Gundersanne: Indeed these are cruft now, but the next bot po-refresh will clean them up. I already imported the translations into cockpit-ostree, so they aren't lost either.