Ahem, sorry. I may have gotten this confused; is this for a trigger branch run or for running infrastructure/ tests?

This is for adding a trigger run of Python 3 on triggers/chrome_dev and triggers/firefox_nightly (item 2 mentioned in your email)

Pushed 3effe467b657cb6a61b1cee55bd5cc5136f02a0a to triggers/chrome_dev. The closest SHA with a run on wpt.fyi is its grandparent, 9a6026305062c90d84a567d81434010dde6c6c22: https://wpt.fyi/results/?run_id=469690003

@ziransun - do I understand correctly that there's no point pushing to triggers/firefox_nightly yet, as Firefox does not yet work in py3?

https://community-tc.services.mozilla.com/tasks/groups/TriD1MleTcOj7RNYgKFZaA should be the TaskCluster URL for the triggers/chrome_dev run

@ziransun - do I understand correctly that there's no point pushing to triggers/firefox_nightly yet, as Firefox does not yet work in py3?

Yes.

Looks like it failed very quickly? One of the logs has:

2020-04-22 19:47:09,848 - tc-run - INFO - Downloading manifest in Python 3 environment
Traceback (most recent call last):
  File "./wpt", line 4, in <module>
    from tools.wpt import wpt
  File "/home/test/web-platform-tests/tools/wpt/wpt.py", line 10, in <module>
    from . import virtualenv
  File "/home/test/web-platform-tests/tools/wpt/virtualenv.py", line 5, in <module>
    from distutils.spawn import find_executable
ModuleNotFoundError: No module named 'distutils.spawn'

Looks that we need to install python3-distutils.

I wonder what happens if we escape manifest download and only do wpt run with python 3

I wonder what happens if we escape manifest download and only do wpt run with python 3

If you look further in the log, I think we hit the same problem. It actually continues even though the manifest call failed!

Can we do "sudo apt-get install python3-distutils" in the linux machine that is running this task?

Can we do "sudo apt-get install python3-distutils" in the linux machine that is running this task?

@Hexcles is this the sort of thing that we would normally do in our Dockerfile, or somewhere else? For this temporary testing, I would also be happy to just install it somewhere hacky if that works.

I think we should just add python3-pip to the base image, which is also required by tox and depends on python3-distutils.

• Add python3 and python3-pip here:

  wpt/tools/docker/Dockerfile

  Lines 26 to 27 in 959a0d0

  	python \
  	python-pip \

• Get a Docker Hub account. Build and push the image.
• Change this to the new image:

  wpt/.taskcluster.yml

  Line 59 in 959a0d0

  image: hexcles/web-platform-tests:0.35

• Drive-by: I think we can remove this install section:

  wpt/tools/ci/tc/tasks/test.yml

  Lines 102 to 107 in 959a0d0

  	tox-python3:
  	env:
  	TOXENV: py36
  	PY_COLORS: 0
  	install:
  	- python3-pip

Add python3 and python3-pip here:

Curious, how do we get python3 today? (Since it seems to be available in TC?)

Curious, how do we get python3 today? (Since it seems to be available in TC?)

Both python and python3 are included in the base ubuntu:18.04 image (so technically we don't need to list either in apt-get install, but it also doesn't hurt).

@stephenmcgruer:
Does it make sense to trigger another run based on the wptserve changes made by @Hexcles? This will give me a better idea of failed tests in Python 3.

Happy to do so; can you rebase this PR to master to pick up the changes @Hexcles has made? I can then push the new SHA to the trigger branches.

I have pushed d47787d (parent commit d929bed), to triggers/chrome_dev.

https://community-tc.services.mozilla.com/tasks/groups/V1c_ockSTQKTwgZTUyCdcQ should be the TaskCluster run for it.

Thank you!

I think https://wpt.fyi/results/?diff&filter=ADC&run_id=528790002&run_id=519310004 should be the diff.

I have pushed 2ce3aa7 (parent commit 2990842), which is a cherry-pick of this change onto master, to triggers/chrome_dev.

https://community-tc.services.mozilla.com/tasks/groups/dc8GrlEaRAKicX-tr97Lgw should be the Taskcluster run for it.

I believe https://wpt.fyi/results/?diff&filter=ADC&run_id=601030002&run_id=629830001 should be the diff

@stephenmcgruer: Is it okay to trigger another run? The PRs that I mentioned during the meeting have now been merged in. Thank you!

I have pushed 926a8ed (parent commit 7c7fe3f), which is a cherry-pick of this change onto master, to triggers/chrome_dev.

https://community-tc.services.mozilla.com/tasks/groups/eW7WDm5CQ36lcfl6pZfnFQ should be the Taskcluster run for it.

I believe https://wpt.fyi/results/?diff&filter=ADC&run_id=632440001&run_id=619000003 is the diff for the two runs.

Awkwardly, chrome_dev updated right in-between the two runs which has caused some churn in results -_-, so I'm going to wait until tomorrow and run another trigger run to get a cleaner comparison.

I have pushed 038f8e8 (parent commit fd5a851), which is a cherry-pick of this change onto master, to triggers/chrome_dev.

https://community-tc.services.mozilla.com/tasks/groups/ZZUrTFXoRQS7DYRuQ1n65Q should be the Taskcluster run for it.

@ziransun I have triggered a Firefox nightly run for 083393c, which is this CL rebased onto master and with a fix for #25012.

https://wpt.fyi/results/?run_id=638560001&run_id=642320001&diff should be the wpt.fyi diff against its nearest parent, please take a look

@ziransun I have triggered a Firefox nightly run for 083393c, which is this CL rebased onto master and with a fix for #25012.

https://wpt.fyi/results/?run_id=638560001&run_id=642320001&diff should be the wpt.fyi diff against its nearest parent, please take a look

@stephenmcgruer: Thanks! I will check on this.

I have pushed 9d65b30 , which is a cherry-pick of this PR onto master, to both triggers/chrome_dev and triggers/firefox_nightly. The parent commit is 6ceb598 .

I will update with wpt.fyi URLs once the runs complete and have results processed.

https://wpt.fyi/results/?diff&filter=ADC&q=is%3Adifferent&run_id=685140004&run_id=687340002

https://wpt.fyi/results/?diff&filter=ADC&q=is%3Adifferent&run_id=685170001&run_id=683210001

Should be the two runs.

https://wpt.fyi/results/?diff&filter=ADC&q=is%3Adifferent&run_id=685140004&run_id=687340002

https://wpt.fyi/results/?diff&filter=ADC&q=is%3Adifferent&run_id=685170001&run_id=683210001

Should be the two runs.

Thank you @stephenmcgruer! I'll look into them.

Closing this PR; we have switched the CI to Py3 and I think this particular experiment is not needed.