Build has FAILED

Patch application report for D4981 (id=17772)

Rebasing onto cf0583b079...

Current branch diff-target is up to date.

Changes applied before test

commit b7ec5ec0096d813265150f8b2195953ac1a55130
Author: Nicolas Dandrimont <nicolas@dandrimont.eu>
Date:   Mon Feb 1 12:37:23 2021 +0100

    journal client: add a test for real-world message processing
    
    This adds a dump of the messages related to linux.git, and checks that
    permutations of these messages yield sensible data in the origin visit
    cache.

Link to build: https://jenkins.softwareheritage.org/job/DSCH/job/tests-on-diff/292/
See console output for more information: https://jenkins.softwareheritage.org/job/DSCH/job/tests-on-diff/292/console

Build is green

Patch application report for D4981 (id=17799)

Rebasing onto aaffff2631...

First, rewinding head to replay your work on top of it...
Applying: journal client: add a test for real-world message processing

Changes applied before test

commit 2d760f3404377e7070aad322f26e0faea4a6179d
Author: Nicolas Dandrimont <nicolas@dandrimont.eu>
Date:   Mon Feb 1 12:37:23 2021 +0100

    journal client: add a test for real-world message processing
    
    This adds a dump of the messages related to linux.git, and checks that
    permutations of these messages yield sensible data in the origin visit
    cache.

See https://jenkins.softwareheritage.org/job/DSCH/job/tests-on-diff/295/ for more details.

Cross-checking out of the db, and agreed on the test's output expectation.

I don't know if we want that test to be included as well.

Also, I'm surprised at the overall tests speed (around 65s only with that test [1]).
Last time we added a test with 720 permutations, it took quite a while to run
(but it was not with hypothesis, plain pytest.mark.parametrize and that was too
slow so we decreased the sample a lot).

Curious me, what does the strategies.permutations do, take a subset out of the
permutations and run the test on it?

[1] 259 passed, 1 xfailed in 66.15s (0:01:06) with that test that ran :)

In D4981#125945, @ardumont wrote:

Also, I'm surprised at the overall tests speed (around 65s only with that test [1]).
Last time we added a test with 720 permutations, it took quite a while to run
(but it was not with hypothesis, plain pytest.mark.parametrize and that was too
slow so we decreased the sample a lot).

Curious me, what does the strategies.permutations do, take a subset out of the
permutations and run the test on it?

Yes, it does the test on a limited random sample of permutations (like all random hypothesis tests, the sample is controlled by the hypothesis seed). If a test fails, hypothesis records the seed and the permutation that failed, and tries to reduce the sample to the smallest number of actual 1-1 permutations of lines.

That probably can be closed since the journal client evolved since the time of this diff.