Open Bug 1429794 Opened 6 years ago Updated 2 years ago

Re-enable dom/workers/test/serviceworkers/isolated/multi-e10s-update/browser_multie10s_update.js

Categories

(Core :: DOM: Service Workers, enhancement, P3)

enhancement

Tracking

()

People

(Reporter: asuth, Unassigned)

References

(Blocks 1 open bug)

Details

As part of bug 1416879 I've been fighting with browser_multie10s_update.js. The story goes:
- browser_force_refresh.js didn't clean up its registration that used the same scope, so the test could end up using the wrong SW for updating, resulting in both tests passing.
- I added an unregister to browser_force_refresh.js
- Then we suffered from the inherent multi-e10s unregister propagation races (the fix which I developed for parent intercept is very ugly), so we moved the e10s-update test into its own subdir, isolated/multi-e10s-update/.
- The test had been made to only run on windows for test failure reasons.
- Now the test is failing to perform both updates on linux stylo debug runs with failures like at https://treeherder.mozilla.org/#/jobs?repo=try&revision=7868dd48ed151ba563e57c8c5ad9db144f51d614 so we're just turning the test off for now because the blocker is very serious and the test is just a distraction at this point.

The failures above include some MOZ_LOGging I added which show that:
- the two update() calls are being issued
- one is being allowed, one is being denied
- the second script request is NOT being issued
- the update that should succeed is failing

nsHttp logging or some additional SW logging would likely shed better light on why the update request is failing.  That's for later and I'll track it down.

The disabling is happening as part of the bug 1416879 landing that cleans this all up, or tries to.
Priority: -- → P2
Backlog this bug?
Flags: needinfo?(bugmail)
Assignee: bugmail → nobody
Status: ASSIGNED → NEW
Flags: needinfo?(bugmail)
Priority: P2 → P3
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.