1081996 - Disable e10s prompt

Status: RESOLVED FIXED

(Testing Graveyard :: Mozmill, defect)

Description

[Andrei Eftimie]

I've seen the e10s prompt interfere with our tests (l10n test noticeably)

We should disable it globally.
The following preference takes care of it.

Preference
----------
Name: `browser.displayedE10SPrompt.1`
Value: `5`

I'd include this in the next mozmill release, we could try disabling this in our tests as a workaround if it disrupts more than right now.

Comment 1

[Henrik Skupin [:whimboo][⌚️UTC+1]]

This was already fixed in Mozmill with release 2.0.7 via bug 1065436. Please check what version of mozmill you are using. It looks like an outdated one.

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Turned out we use a wrong pref in Mozmill, which needs a fix. Waiting for the reply from Jimm on bug 1065436 before we can get started here. But yeah, looks like we need a 2.0.9 release of Mozmill.

Comment 3

[Henrik Skupin [:whimboo][⌚️UTC+1]]

(In reply to Jim Mathies [:jimm] from bug 1065436 comment #8)
> > +            # Bug 695026 - Re-enable e10s when fully supported
> > +            preferences.append(('browser.tabs.remote', False))
> > +            preferences.append(('browser.tabs.autostart', False))
> 
> This should be 'browser.tabs.remote.autostart'. Also 'browser.tabs.remote'
> was depreciated about a week ago.

After talking to Jim, I will remove the remote pref, and update the autostart. Thanks for noticing both of you!

Comment 4

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Attachment: Finally disable v1

Strangely I'm not able to reproduce this problem at all on my Linux box. Whether with Mozmill directly nor the automation scripts. So we will have to wait for Andrei to verify that it completely fixes the problem.

Comment 5

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Comment on attachment 8504196 [details] [diff] [review]
Finally disable v1

Review of attachment 8504196 [details] [diff] [review]:
-----------------------------------------------------------------

Today is Canadian holiday. So lets move the review request over to Chris.

Comment 6

[Jim Mathies [:jimm]]

You might want to take a look at this code to be sure everything is covered - 

http://mxr.mozilla.org/mozilla-central/source/toolkit/xre/nsAppRunner.cpp#4582

I wonder about that trial pref 'browser.tabs.remote.autostart.1' and subsequent flags.

Comment 7

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Would you mind telling for what this additional pref is used? All the flags for e10s mode are way too confusing. Things have changed a lot, and I currently do not have the time to dive into all that Firefox code. I would appreciate to know if we have to also add this trial pref to Mozmill.

Comment 8

[Jim Mathies [:jimm]]

lets ask felipe, he knows more about that .1 pref than I do.

Comment 9

[:Felipe Gomes (needinfo for replies!)]

The prefs are confusing but we've been simplifying them. Basically there are 3 prefs now:

browser.tabs.remote.autostart
-----------------------------
meant for users to manually toggle (through the "enable e10s" checkbox Preferences dialog) when they want to opt-in to e10s testing

browser.tabs.remote.autostart.X
-------------------------------
not used yet, but will be used when we want to force a period of e10s on nightly builds for a few days

browser.displayedE10SPrompt.X
-----------------------------
will be used (and X incremented) each time we have a period where we want to offer the e10s prompt asking users to opt-in

So, in order to both disable the prompt, and e10s itself, and avoid having to deal with the same bug in the future, I suggest that you set all of:

browser.tabs.remote.autostart = false

browser.tabs.remote.autostart.1 = false
browser.tabs.remote.autostart.2 = false
browser.tabs.remote.autostart.3 = false
browser.tabs.remote.autostart.4 = false

browser.displayedE10SPrompt.1 = 5
browser.displayedE10SPrompt.2 = 5
browser.displayedE10SPrompt.3 = 5
browser.displayedE10SPrompt.4 = 5

And then remove these when bug 695026 gets fixed.

Comment 10

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Thanks Felipe. Can we be sure we only have indices from 1 to 4, or should we account more?

Comment 11

[:Felipe Gomes (needinfo for replies!)]

Not sure if there will be more or not, I guess it depends on how the experiments with Nightly users go. I guess you can add more as it's trivial to do so, but too much would be unnecessary. I'll keep an eye and if we ever end up going higher again I can ping you or open a new bug to add more.

Comment 12

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Attachment: Finally disable v1.1

Ok, this adds the additional prefs and even some more (up to .9). That should be enough and we may not have to release another 2.0.x release of Mozmill.

Comment 13

[Andrei Eftimie]

Comment on attachment 8504374 [details] [diff] [review]
Finally disable v1.1

Review of attachment 8504374 [details] [diff] [review]:
-----------------------------------------------------------------

Works fine.

Comment 14

[Henrik Skupin [:whimboo][⌚️UTC+1]]

https://github.com/mozilla/mozmill/commit/0d72d6e5195baed07636b907b557c82d05854bb4 (master)
https://github.com/mozilla/mozmill/commit/c2012e6a5cf817e4fb6a6a1675119e69244ded8d (hotfix-2.0)