1010999 - Change update notification to Weekly instead of Daily on Aurora

Status: RESOLVED FIXED

(Firefox :: General, defect)

Description

[Rob Campbell [:rc] (:robcee)]

It would be nice if Aurora builds updated weekly instead of daily. They could (and should) still download the daily fixes but we shouldn't prompt users to update for a longer period.

This is a request for Developer Tools users we push to Aurora for a "more stable" experience than Nightly. The perception with the daily update is that it's just as churny as Nightly and we'd like to mitigate that somewhat.

Comment 1

[Robert Strong (they/them - no direct email)]

The in product check is controlled by application preferences so each product can choose how often to check and whether an update is created is a app / product driver decision (e.g. just as Firefox beta updates are created weekly) with the work performed by releng... so, neither of those things would be done by the app update client code.

Moving to releng and cc'ing Lawrence. I suggest you also send email to release-drivers to start a discussion for this change.

Comment 2

[Rail Aliiev [:rail]]

Lukas,

do we have any resolution here?

Comment 3

[Lukas Blakk [:lsblakk] use ?needinfo]

As no discussion was started on release-drivers we should do that first to ensure there are no issues we haven't considered.  Will send that now.

Comment 4

[:Gavin Sharp [email: [e-mail]]]

There was a discussion started on release-drivers ("Proposal to change update nag frequency to 1 week on Aurora"), it just didn't reach a useful conclusion.

I think there are no significant objections to changing the nag frequency, which as I understand it is an in-product change.

There was some objections to changing the update frequency, so let's drop that.

Comment 5

[Jeff Griffiths (:canuckistani) (:⚡︎)]

(In reply to :Gavin Sharp (email [e-mail]) from comment #4)
> There was a discussion started on release-drivers ("Proposal to change
> update nag frequency to 1 week on Aurora"), it just didn't reach a useful
> conclusion.
> 
> I think there are no significant objections to changing the nag frequency,
> which as I understand it is an in-product change.

Would love this. Would it also extend to Android?

Comment 6

[Robert Strong (they/them - no direct email)]

(In reply to Jeff Griffiths (:canuckistani) from comment #5)
>...
> Would love this. Would it also extend to Android?
This bug would change it for desktop only. File an Android bug and start a discussion with the Android drivers if you'd like it to be changed there as well.

Comment 7

[Robert Strong (they/them - no direct email)]

Gavin, Kevin brings up a good point in the release-drivers discussion in that the vast majority of people restart anyways after two days and suggests going with a 2 day nag to restart just as we have on beta and release. Do you think it should be changed to two days and if it is determined that a longer period would be valuable it can be done another bug?

Comment 8

[Robert Strong (they/them - no direct email)]

Attachment: patch implemented per comment #7

Gavin, this implements what I outlined in comment #7 in case you agree with its conclusion.

Comment 9

[:Gavin Sharp [email: [e-mail]]]

(In reply to Robert Strong [:rstrong] (use needinfo to contact me) from comment #7)
> Gavin, Kevin brings up a good point in the release-drivers discussion in
> that the vast majority of people restart anyways after two days

Where does this conclusion come from?

I think the goal is to avoid annoying the minority of people who leave their browsers open pretty much always (≃ developers), so I think I would prefer to go with the originally proposed 1 week.

Comment 10

[Robert Strong (they/them - no direct email)]

(In reply to :Gavin Sharp (email [e-mail]) from comment #9)
> (In reply to Robert Strong [:rstrong] (use needinfo to contact me) from
> comment #7)
> > Gavin, Kevin brings up a good point in the release-drivers discussion in
> > that the vast majority of people restart anyways after two days
> 
> Where does this conclusion come from?
Initially from crash data (which I was involved in gathering) and then from a test pilot study (which I was not involved in gathering).

> I think the goal is to avoid annoying the minority of people who leave their
> browsers open pretty much always (≃ developers), so I think I would prefer
> to go with the originally proposed 1 week.
No problem

Comment 11

[Robert Strong (they/them - no direct email)]

Attachment: patch - notify after 7 days

Comment 12

[Justin Wood (:Callek)]

(In reply to :Gavin Sharp (email [e-mail]) from comment #9)
> (In reply to Robert Strong [:rstrong] (use needinfo to contact me) from
> comment #7)
> I think the goal is to avoid annoying the minority of people who leave their
> browsers open pretty much always (≃ developers), so I think I would prefer
> to go with the originally proposed 1 week.

I think product developers can be expected to be able to tweak the pref if it annoys them, no?

> release-drivers discussion ... and suggests going with a 2 day nag to restart just as we have on beta and release.

Is there a reason you're discounting the resolution of that discussion and going with a full week nag?

(I'm not a product dev, and can change this as well, but I find it quite disingenuous if we nag more often on release/beta when aurora changes more often. And we actually _want_ our devs dogfooding, right?)

Comment 13

[Robert Strong (they/them - no direct email)]

(In reply to Justin Wood (:Callek) from comment #12)
> (In reply to :Gavin Sharp (email [e-mail]) from comment #9)
> > (In reply to Robert Strong [:rstrong] (use needinfo to contact me) from
> > comment #7)
> > I think the goal is to avoid annoying the minority of people who leave their
> > browsers open pretty much always (≃ developers), so I think I would prefer
> > to go with the originally proposed 1 week.
> 
> I think product developers can be expected to be able to tweak the pref if
> it annoys them, no?
> 
> > release-drivers discussion ... and suggests going with a 2 day nag to restart just as we have on beta and release.
> 
> Is there a reason you're discounting the resolution of that discussion and
> going with a full week nag?
> 
> (I'm not a product dev, and can change this as well, but I find it quite
> disingenuous if we nag more often on release/beta when aurora changes more
> often. And we actually _want_ our devs dogfooding, right?)
I don't see it that way since release and beta get fewer updates the nag already happens less often so the 2 days before a nag turns into 9 days before a nag if we have an update once a week and turns into 6 weeks and 2 days before a nag if we release every 6 weeks.

Also, I have been told that test pilot data shows the vast majority of the time (iirc 98.5%) Firefox is restarted anyways within 48 hours so this shouldn't have a significant impact on crashstats, etc.

Comment 14

[Chris Peterson [:cpeterson]]

(In reply to Robert Strong [:rstrong] (use needinfo to contact me) from comment #13)
> Also, I have been told that test pilot data shows the vast majority of the
> time (iirc 98.5%) Firefox is restarted anyways within 48 hours so this
> shouldn't have a significant impact on crashstats, etc.

Do we have test pilot data for how often users restart for each release channel?

Comment 15

[Robert Strong (they/them - no direct email)]

I don't know since I wasn't involved in the gathering of the data or the generation of the report. Also, it would be less applicable on Aurora due to the update cadence.

Comment 16

[:Gavin Sharp [email: [e-mail]]]

(In reply to Justin Wood (:Callek) from comment #12)
> I think product developers

Web developers, not product developers. Or "power users"/enthusiasts if you prefer.

> Is there a reason you're discounting the resolution of that discussion and
> going with a full week nag?

There was no "resolution of that discussion"; Kevin made a proposal that I disagree with. There's no reason that the update nag needs to be progressively longer across channels. If the majority of users restart after 2 days anyhow, this won't have a significant impact on uptake rate, and if it does we can always revisit.

Comment 17

[Robert Strong (they/them - no direct email)]

Pushed to fx-team
https://hg.mozilla.org/integration/fx-team/rev/45fd2bac9153

in-testsuite- though we do have tests for the nag notification for short intervals to actually test it would require running the test for a week. ;)

Comment 18

[Benjamin Smedberg]

FWIW, I can collect data about average session length on the various channels when I'm back next week. I expect to have a trimodal distribution:

* people who launch and close the browser all the time (several times a day)
* people who have their browser open all day but close it at night
* people who typically leave their browser open all the time

I'll post results to firefox-dev next week, but I expect that we're still better off if we only nag the third group once a week. Perhaps we should consider a less invasive form of the update prompt on a shorter timeframe.

Comment 19

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/45fd2bac9153

Comment 20

[:Gavin Sharp [email: [e-mail]]]

Comment on attachment 8435348 [details] [diff] [review]
patch - notify after 7 days

Let's uplift to Aurora, since this is only effective there.

Comment 21

[Robert Strong (they/them - no direct email)]

This landed before the merge to aurora.

Comment 22

[Robert Strong (they/them - no direct email)]

Comment on attachment 8435348 [details] [diff] [review]
patch - notify after 7 days

http://mxr.mozilla.org/mozilla-aurora/source/browser/branding/aurora/pref/firefox-branding.js#15

Comment 23

[Rob Campbell [:rc] (:robcee)]

Thanks everyone!