Closed Bug 1147156 Opened 9 years ago Closed 9 years ago

Re-enable e10s by default for Nightly users who may have disabled it

Categories

(Firefox :: General, defect)

x86
Windows 7
defect
Not set
normal
Points:
2

Tracking

()

RESOLVED FIXED
Firefox 40
Iteration:
40.1 - 13 Apr
Tracking Status
e10s m6+ ---
firefox40 --- fixed

People

(Reporter: mconley, Assigned: Felipe)

References

Details

Attachments

(3 files)

According to jimm, we have 78% of our Windows and Mac Nightly population running with e10s enabled by default, and about 60% of our Linux users.

We might want to consider flipping it on (again) by default for our Nightly population to see if:

1) Issues that were causing users to turn it off have been resolved, thereby allowing them to continue to use it
2) Issues that we have not yet filed or found good STR for are affecting the usability of the browser that would prevent an uplift.
From #e10s:

"17:42: * jld makes mental note to turn it back off *again*. Sigh. (Bug 1058542)"

Bug 1058542 is the bug for NoScript e10s compat.

Perhaps we can also be less hamfisted about this, and enable for users that don't have one of our known-busted-add-ons enabled?
Waiting until 40a1 seems better timing. Displaying the uplift intention might get some extra feedback.

The password add-ons are a big no-go for some.

From my reading of telemetry graphs about 2/3 of windows users not running e10s are blocked from running rather than disabled by choice. (25% overall for windows blocked, ~65% running e10s)
Assignee: nobody → mconley
So, we discussed this in the e10s triage meeting last week, and the consensus was "yes, reactivating this across our Nightly population might piss off our Nightly users. But that's what Nightly is for."

How do you suggest we do this, felipe? We did this kinda accidentally in the past by toggling off the e10s pref for a few days, and then toggling it back on - which seemed to have the byproduct of flipping the pref for people who'd had it disabled. I imagine we probably want to do something simpler than landing two patches several days apart... should we bump all usage of browser.tabs.remote.autostart.1 to browser.tabs.remote.autostart.2?
Flags: needinfo?(felipc)
Yeah, I think that's the way to go. At the same time, we should reset .1 to false so we can re-use it in the future!

Some bug archeology might be needed because if I recall there were uses of browser.tabs.remote.autostart.1 outside of m-c, for some automation tools.
Flags: needinfo?(felipc)
The bugs I was thinking of were bug 1081996, bug 1082647, bug 1093691

They all covered .autostart.2, so we're good to go.

two things to keep in mind: bug 1093691 only added up to .autostart.2, so if we ever need .3 we will need to handle it.

And browser.displayedE10SPrompt.N was meant to be paired with .autostart.N, but I don't really think it's necessary.
Assignee: mconley → felipc
We will be using browser.tabs.remote.autostart.2 now, but it's good to keep .1 around being set to false because we might reuse it.
Attachment #8587555 - Flags: review?(poirot.alex)
Attachment #8587555 - Flags: review?(fabrice)
Here's something extra that I wanted to do. Currently there's no way for us to reshow the e10s notice ("You're now helping to test multi-process..") _only_ to the users who are being re-enabled. We can just show it again to everyone on e10s if we bump CURRENT_NOTICE_COUNT.

With this, the notice will show itself again to anyone who voluntarily had disabled e10s and is now being re-enabled (including by the user toggling the checkbox again)

This will only take effect on the next time we re-enable users.
Attachment #8587557 - Flags: review?(mconley)
Attachment #8587555 - Flags: review?(poirot.alex) → review+
Attachment #8587555 - Flags: review?(fabrice)
Comment on attachment 8587556 [details] [diff] [review]
bump autostart.1 to .2, browser files

Do we need to add autostart.2 here as well?: https://dxr.mozilla.org/mozilla-central/source/testing/profiles/prefs_general.js#306
Attachment #8587556 - Flags: review?(mconley) → review+
Comment on attachment 8587557 [details] [diff] [review]
Extra - improve notice handling

Review of attachment 8587557 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks felipe!
Attachment #8587557 - Flags: review?(mconley) → review+
(In reply to Mike Conley (:mconley) - Needinfo me! from comment #9)
> Comment on attachment 8587556 [details] [diff] [review]
> bump autostart.1 to .2, browser files
> 
> Do we need to add autostart.2 here as well?:
> https://dxr.mozilla.org/mozilla-central/source/testing/profiles/
> prefs_general.js#306

ah, yes
https://hg.mozilla.org/mozilla-central/rev/153904f1e64a
https://hg.mozilla.org/mozilla-central/rev/820d39343462
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
Target Milestone: --- → Firefox 40
Depends on: 1152002
Blew away my session :(

I got errors from session restore and put at https://pastebin.mozilla.org/8829148

Only disabling e10s, replacing sessionstore.js, and starting again allowed session to be restored again.
Points: --- → 2
Flags: qe-verify-
Flags: firefox-backlog+
Iteration: --- → 40.1 - 13 Apr
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: