Support for Http3 WebTransport server for WPT tests
Categories
(Firefox Build System :: Toolchains, enhancement)
Tracking
(firefox113 fixed)
Tracking | Status | |
---|---|---|
firefox113 | --- | fixed |
People
(Reporter: jesup, Assigned: ahochheiden)
References
(Blocks 1 open bug)
Details
Attachments
(1 file)
We need to support the equivalent to --enable-webtransport-h3 for CI tests of testing/web-platform/tests/webtransport
See also bug 1806698
WPTs are the primary way we're testing WebTransport, so this is important
Reporter | ||
Updated•2 years ago
|
Comment 1•2 years ago
•
|
||
Edit: Sorry about that, I misunderstood what this revision was actually doing.
Updated•2 years ago
|
Assignee | ||
Comment 3•2 years ago
|
||
All relevant dependencies should now be available on our internal PyPi mirror: https://pypi.pub.build.mozilla.org/pub/
See:
https://bugzilla.mozilla.org/show_bug.cgi?id=1820065
https://bugzilla.mozilla.org/show_bug.cgi?id=1820032
https://bugzilla.mozilla.org/show_bug.cgi?id=1820488
Updated•2 years ago
|
Assignee | ||
Updated•2 years ago
|
Comment 4•2 years ago
|
||
Are we likely to end up in a state where if we bump the versions of these dependencies in upstream wpt we're going to run into an error when trying to sync those changes? We try very hard to minimise the amount of manual work required to keep wpt in sync, because even with the minimal manual work we can get away with for two-way sync (resolving conflicts), we still end up with a lot of overhead.
If that is true, and it's the best we can manage for now, I'll need to think about whether there's a way to avoid taking automatic version bumps upstream as an interim solution. But it doesn't seem like a long-term sustainable setup.
Assignee | ||
Comment 5•2 years ago
•
|
||
Yes, but isn't that unavoidable? Assuming we have additional dependencies that upstream does not have, any time upstream bumps versions of their dependencies, there's a chance for version conflicts.
I'm not sure if there's anything we can do to mitigate potential dependency version conflicts, other than minimizing what we depend on, and in this particular case we must depend on these packages, or we can't test webtransport, no?
Comment 6•2 years ago
|
||
I don't think it's about conflicts, as such, just about the fact that we require manual work to update our PyPI mirror. If, for example, we could fetch dependencies from PyPI (which obviously has issue with availability), or a system where we automatically update our PyPI mirror with upstream packages when they're first used, then it wouldn't be a problem.
Usually we work around this by vendoring stuff (with some tradeoff), but that doesn't work well for binary packages.
Comment 8•2 years ago
|
||
bugherder |
Description
•