-
Notes archive: https://github.com/web-platform-tests/wpt-notes/tree/master/minutes
-
Infra issues: https://github.com/web-platform-tests/wpt/issues?q=is%3Aopen+label%3Ainfra+-label%3Adependencies
Attendees: Weizhong Xia, Panos Astithas, Sam Sneddon, James Graham, James Scott
- jgraham: Have not had any runs for awhile. Maybe related to Python 3.11. The error is obscure though. Getting the wrong arguments into subprocess.popen.
File "D:\a\1\s\tools\wptrunner\wptrunner\testrunner.py", line 210, in init
self.browser.start(group_metadata=group_metadata, **self.browser_settings)
File "D:\a\1\s\tools\wptrunner\wptrunner\browsers\base.py", line 318, in start
self._run_server(group_metadata, **kwargs)
File "D:\a\1\s\tools\wptrunner\wptrunner\browsers\base.py", line 341, in _run_server
self._proc.run()
File "d:\a\1\s\_venv3\lib\site-packages\mozprocess\processhandler.py", line 946, in run
self.proc = self.Process([self.cmd] + self.args, **args)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "d:\a\1\s\_venv3\lib\site-packages\mozprocess\processhandler.py", line 160, in __init__
subprocess.Popen.__init__(self, args, **kwargs)
File "C:\hostedtoolcache\windows\Python\3.11.4\x64\Lib\subprocess.py", line 1026, in __init__
self._execute_child(args, executable, preexec_fn, close_fds,
File "d:\a\1\s\_venv3\lib\site-packages\mozprocess\processhandler.py", line 295, in _execute_child
(
ValueError: too many values to unpack (expected 22)
- sam: do we block interop score on edge?
- jgraham: breaks the home page but not the interop score. Maybe not as urgent.
- sam: Do we run the infra test against edge on windows? It would have been nice to have catch this before landing the PR.
- jgraham: upon checking, only seeing the unit tests not the infra tests. We could enable that job.
- sam: For most of the tooling changes, we are careful.
Next steps:
- jgraham: Investigate the failure in more detail. Consider reverting (only if necessary)
- Add infrastructure job for Windows
RFC 157 - Add feature meta tag to web-platform-tests meta data
- james: WebDX has the feature-set repo. Would be interesting to be able to filter by feature as well as by spec. Naming is up for discussion. As feature-set comes online it would allow people to corelate that project with wpt. Works similarly to spec manifest.
- jgraham: historically, we have been cautious of adding metadata. Adding optional metadata is probably fine. But then who is responsible for keeping it up to date. Will be opposed to anything mandatory.
- james: I imagine we'd need some kind of monitoring system in place, so that WebDX can see changes. That could come as we try it out. Agree it can be optional. RFC should also document how to remove the metadata system if it doesn't work out. I'll add that to the RFC. In terms of who's responsible, I'll work on adding some initial metadata then we can experiment as the feature set repo gets more data.
- Panos: BCD hasn't failed yet, so maybe that's a reason to be optimisitc if it's the same people.
Action items:
- jcscottiii: Add note to RFC about to revert and who will be doing initial contributions
WPT Issue 40952 - Can not run base test alone when a test file has variants
- Weizhong: Discussion and do a status update for subsuite support
- Quick update: Thanks to jgraham. Can now run subsuite tests locally. Now facing an issue regarding the total number of tests doesn't match
- Next: upload patch to upstream. Should I upload separately or push on top of existing
- jgraham: It's okay to do either.
- weizhong: will push on top of the existing one. Some tests are run multiple shards . Due to the some variants having an empty string.
- sam: I wonder if there's some difference between passing a path or what is basically a partial URL
- jgraham: looks for tests at the path. Then it appends a forward slash and treats it as an ID.
- sam: what's the suggested behavior
- jgraham: Either options require RFCs. Options are 1) having a non-zero length variant (requires manifest and tests to be update) or 2) in wptrunner adding a leading slash is interpreted as the exact test ID and never match any variants. No strong opinions on which one
- weizhong: Sounds like the first one is easier. Could we start with linting
- sam: probably needs a lint
Next steps:
- weizhong: Add changes on top of jgraham's changes
- weizhong: add a RFC for option 1.
WPT Issue 40860 - WebDriver BiDi: Expand list of people with admin merge permissions
- panos: Pulse check on the idea. And also when was the last audit of admins
- sam: Have not done audit in years. Concern is not about adding more admins
- jgraham: Either making people on the core team or actively contributing to the infra. There's two problems: The chrome stability job and the firefox job only once. Preference is to solve the issues than to add more admins to force merge. Next agenda item to audit the admins.
Next steps:
- jgraham: fix the firefox issue
- next meeting talk about admin to time zone coverage
Other Notes:
- Safari Runs on Azure
- jgraham escalated the issue. Waiting to hear back.
- Contact points for each browser Link
- Previous Discussion
- jcscottiii: No progress on my end.