Allow overriding all preferences (eg. media.autoplay.default) for testing
Categories
(GeckoView :: Media, defect, P3)
Tracking
(Not tracked)
People
(Reporter: whimboo, Unassigned)
References
(Blocks 2 open bugs)
Details
Attachments
(1 obsolete file)
For the upcoming streaming tests for Raptor I need the media.autoplay.default
preference to be set to 0
by default, so that the test like [1] will be started right after loading the page.
This works all fine with Firefox but fails with any Android based browser.
As James already mentioned we override some preferences via GeckoRuntimeSettings, and that we also get rid of user defined preferences. :/
I hope that we could find a way to enable autoplay, because that is somewhat a blocker for me to get those Raptor tests (bug 1539111) implemented.
[1] http://yttest.dev.mozaws.net/2019/main.html?test_type=playbackperf-test&command=run:8
Yeah, I guess we could just not reset pref values for "managed" prefs. https://searchfox.org/mozilla-central/source/mobile/android/components/geckoview/GeckoViewStartup.js#162
Comment 2•5 years ago
|
||
I'm pretty concerned that GV is blazing a "brave new path" for Gecko that diverges from the expectation Gecko developers have without capturing major points of divergence. Managing prefs bit me with tracking protection in Fenix and bit Henrik with media autoplay. This is only going to get worse.
snorp: who is the right person to own documenting GeckoView's preference policy? We can't figure out the right balance between the engine's desire for consistency and control and automation's desire for overrides until we know what is the long-term plan for the engine and what the situation is right now.
This documentation should be sibling to https://mozilla.github.io/geckoview/tutorials/automation.html.
This interacts badly with custom profiles used during tests that set things like
'media.autoplay.default'.
GV is just trying to make GeckoRuntimeSettings
the source of truth here for the prefs that it manages. It's probably being a little too aggressive by unsetting user prefs, though, so the attached patch drops that.
Reporter | ||
Comment 5•5 years ago
|
||
James, please note that there are a couple of linter and test failures for the try push from yesterday. Can you please have a look? Thanks
https://treeherder.mozilla.org/#/jobs?repo=try&revision=ce1d8db9229932349682979d1a0a42cb39aaeefa
Reporter | ||
Comment 6•5 years ago
|
||
Hi James, do you have any kind of update here? Thanks.
Sorry, I've been out for a bit -- I hope to get back to this soon. I'm (mostly) ok with the approach I have in the patch here, but it turns out we were relying on the old behavior for a few junit tests. I need to update the harnesses to not set prefs that GV manages.
Comment 8•5 years ago
|
||
I'm editing a bunch of GeckoView bugs. If you'd like to filter all this bugmail, search and destroy emails containing this UUID:
e88a5094-0fc0-4b7c-b7c5-aef00a11dbc9
Updated•5 years ago
|
Updated•5 years ago
|
Updated•4 years ago
|
Comment 9•4 years ago
|
||
Now that we have the yml
config file we can just detect prefs in that file that are "controlled" by Gecko and override the app's choice in GeckoRuntimeSettings instead. Since the config file is only used for testing this is safe and OK to do.
Updated•4 years ago
|
Updated•3 years ago
|
Updated•3 years ago
|
Updated•3 years ago
|
Updated•3 years ago
|
Comment 11•3 years ago
|
||
(In reply to Emily Toop (:fluffyemily) from comment #10)
Do we still need this Nick?
Need? No, I think not -- the status quo has remained for a very long time at this point. But I do think preference control does come up when testing pretty consistently, so it's worth keeping this open just to collect discussion, I think.
Comment 12•2 years ago
|
||
Moving some media bugs to the new GeckoView::Media component.
Updated•2 years ago
|
Updated•2 years ago
|
Description
•