Ship Windows Runtime version of Geolocation backend
Categories
(Core :: DOM: Geolocation, enhancement)
Tracking
()
People
(Reporter: m_kato, Assigned: m_kato)
References
(Blocks 1 open bug)
Details
Attachments
(1 external file)
I have landed Windows Runtime version of Geolocation backend by bug 1980653 and have turned on Nightly channel. We should ship it on release channel if no problem.
Comment 1•9 months ago
|
||
Do we want to ship this in 149? Bug 1980653 has been on Nightly for about a month and will we get value from having it be Nightly-only for another cycle? Or is it best to let it go to Beta now? I don't see any reports of regressions related to bug 1980653.
Comment 2•9 months ago
|
||
It's a bit late in the beta cycle for uplift but I agree we should fix this ASAP. I assume the next beta is a release candidate, so uplift is very aggressive. Since the Nightly population is so limited, I'm not so convinced this is safe. There is nothing in crash-pings, and I mean nothing (from the new or legacy versions), which could be great or could indicate a problem reporting crashes from that utility process. OTOH, I think this is considered super-high-priority, so maybe we roll the dice.
Comment 3•9 months ago
•
|
||
Do we want to ship this in 149?
I assume the next beta is a release candidate, so uplift is very aggressive.
I think 149 is still our current Nightly? We could prepare the patch, land early in 150 and uplift early to 149. Meaning we'd have a whole beta cycle to see problems. I do not think that more time on Nightly only is likely to make jump out issues? Given the patch is just a pref flip, we could even land it now, still on 149, which seems the cleanest solution.
or could indicate a problem reporting crashes from that utility process
In general we are definitely seeing crash reports coming from the utility process, so I would not worry about that part.
And WindowsRuntimeLocationChild added failure telemetry. So we should be able to see any problems from that? I think we can also be pretty sure that SendFailureTelemetry works in general, as we see some?
Comment 4•9 months ago
|
||
(In reply to David Parks [:handyman] from comment #2)
It's a bit late in the beta cycle for uplift but I agree we should fix this ASAP. I assume the next beta is a release candidate
Don't assume, consult our calendar :)
https://whattrainisitnow.com/release/?version=148
Release Candidate was built and shipped to our beta population 2 days ago. We are preparing GA now. So on the 148 cycle your likely target now is the planned dot release.
(In reply to Jens Stutte [:jstutte] from comment #3)
Do we want to ship this in 149?
I assume the next beta is a release candidate, so uplift is very aggressive.
I think 149 is still our current Nightly? We could prepare the patch, land early in 150 and uplift early to 149. Meaning we'd have a whole beta cycle to see problems.
We merge on Monday, if this is just a pref flip, I would prefer that we land it before the merge to catch any CI issue before we are in beta, this would minimize the work as well as we wouldn't have to uplift.
| Assignee | ||
Comment 5•9 months ago
|
||
I don't see bugs by windows runtime backend. It is the time to turn on it on 150.
Comment 6•9 months ago
|
||
(In reply to Makoto Kato [:m_kato] from comment #5)
I don't see bugs by windows runtime backend. It is the time to turn on it on 150.
Are you saying you don't think we should flip the pref for 149?
Comment 7•9 months ago
|
||
(In reply to Pascal Chevrel:pascalc from comment #4)
We merge on Monday, if this is just a pref flip, I would prefer that we land it before the merge to catch any CI issue before we are in beta, this would minimize the work as well as we wouldn't have to uplift.
Just to clarify, the pref is active on current Nightly, I assume you want to avoid a switch back when merging to beta to then switch it on again after a few days, which indeed feels like useless noise.
Comment 8•9 months ago
|
||
In general we are definitely seeing crash reports coming from the utility process, so I would not worry about that part.
To add another data point, we have general telemetry about utility process launch failures. We had a spike in that telemetry for Nightly in Oct/Nov, but since then it looks pretty flat.
If it's ok on nightly I think it's fair to let it ride the train, and if something urgent happens we can use nimbus to flip it back from stable without a dot release (if dot release can't happen soon enough).
| Assignee | ||
Comment 10•9 months ago
|
||
(In reply to Jens Stutte [:jstutte] from comment #6)
(In reply to Makoto Kato [:m_kato] from comment #5)
I don't see bugs by windows runtime backend. It is the time to turn on it on 150.
Are you saying you don't think we should flip the pref for 149?
149 is soft code freeze now. If a+ from release team, it is OK on 149.
| Assignee | ||
Comment 11•9 months ago
|
||
Ah, I miss Ryan's mail. it is OK on 149.
Comment 12•9 months ago
|
||
Thanks, let's just prepare the patch and land then.
149 is soft code freeze now.
IIUC we dropped soft-freeze, only the string-freeze is left.
| Assignee | ||
Comment 13•9 months ago
|
||
Updated•9 months ago
|
Comment 14•8 months ago
|
||
Can we land this before the merge? Thank you!
Updated•8 months ago
|
Comment 15•8 months ago
|
||
Yeah, I assume he just forgot.
Comment 16•8 months ago
|
||
Comment 17•8 months ago
|
||
| bugherder | ||
Updated•8 months ago
|
Updated•8 months ago
|
Updated•8 months ago
|
Description
•