Add crashtest to be sure we fire pagehide event in documents with sync XHR during pageload
Categories
(Core :: DOM: Core & HTML, enhancement, P2)
Tracking
()
Tracking | Status | |
---|---|---|
firefox68 | --- | fixed |
People
(Reporter: dholbert, Assigned: dholbert)
Details
Attachments
(1 file, 1 obsolete file)
My initial patch in bug 1535388 would've prevented us from firing pageshow if there was a sync XHR during pageload.
It nearly passed on Try (aside from one arcane failure in test_disableScript.xul), so let's add some crashtests that more directly test that pageshow is indeed fired which would've caught my mistake.
Assignee | ||
Comment 1•6 years ago
|
||
Updated•6 years ago
|
Updated•6 years ago
|
Updated•6 years ago
|
Assignee | ||
Comment 2•6 years ago
|
||
My "pagehide" variant doesn't seem to pass for some reason, actually, so I'll just add the pageshow one for now. That one reliably passes, locally, with a mozilla-central debug build.
(Updated patch)
Assignee | ||
Comment 3•6 years ago
|
||
Try run to exercise this test on all platforms:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=725f99eea2ed38e7b40191c34a740dd1b873f116
Assignee | ||
Comment 4•6 years ago
•
|
||
Apparently this event isn't reliably fired in this scenario, even in normal mozilla-central. My crashtest was passing reliably yesterday, locally, and it passes on Android (it's shown as passing in the log for Android 4.3 opt C1 and debug C1 in my try run above) -- but it failed on all other Try platforms (specifically, it never fired pageshow, so it timed out).
I tried to rewrite the test with the sync XHR + navigation happening in an iframe, but that doesn't seem to help -- that still makes pageshow fire quite rarely (with "###!!! ASSERTION: Destroying a currently-showing document" being spammed most of the time).
I'm experimenting with an iframe-based pagehide test now, though, which seems more reliable locally (despite comment 2).
If that works, I might just settle for that at the moment, and hopefully we can extend it to cover pageshow as well once bug 1535388 or another followup is addressed.
Updated•6 years ago
|
Assignee | ||
Comment 5•6 years ago
|
||
Assignee | ||
Comment 6•6 years ago
|
||
Here's a Try run with a slightly older but functionally-equivalent version of this crashtest:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=0320684141902b2311e41e5747aef8227d79aa29
(That try run is layered on top of bug 1537588's patch -- without that patch, this crashtest causes an assertion-failure "Destroying a currently-showing document" as noted in comment 4, so we need that patch before we can land this crashtest.)
Assignee | ||
Updated•6 years ago
|
Comment 8•6 years ago
|
||
bugherder |
Comment 9•6 years ago
|
||
Description
•