1051223 - Updater installs Mozilla Maintenance Service without consent

Status: RESOLVED FIXED

(Toolkit :: Application Update, defect)

Description

[Debeet]

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/36.0.1985.125 Safari/537.36

Steps to reproduce:

I installed Firefox without Mozilla Maintenance Service.


Actual results:

When an update for Firefox came and updater ran it installed Mozilla Maintenance Service.


Expected results:

Updater shouldn't install Mozilla Maintenance Service when I didn't choose it in the installer for Firefox.

Comment 1

[(mostly gone) XtC4UaLL [:xtc4uall]]

Is "Use a background service to install updates" in Options > Advanced > Update also checked?

Comment 2

[Debeet]

I have no such option.

Comment 3

[Debeet]

What I mean by that is that option is not visible at all.

Comment 4

[(mostly gone) XtC4UaLL [:xtc4uall]]

What is the value (true/false) of "app.update.service.enabled" when you search for it in about:config?

Comment 5

[Debeet]

It's "true". I forgot to add that I've uninstalled Mozilla Maintenance Service when the updater installed it.

Comment 7

[Brian R. Bondy [:bbondy]]

Did you install with the stub installer or the full installer originally?

I believe that we set an entry in the registry when it is specifically checked off on install, and on update it won't install it if that registry entry is set.
Perhaps this entry isn't being set correctly for the stub installer or something like that though.

Comment 8

[Debeet]

I installed Firefox through stub installer with unchecking the Mozilla Maintenance Service.

Comment 9

[Brian R. Bondy [:bbondy]]

Attachment: Patch rev1

This sets the "Attempted" reg value so that new upgrades won't try to install the service.  Tested locally and it's working.
Pave over installs with checkbox on will still install it.

Comment 10

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

Did you figure out what changed to make this happen?

Comment 11

[Brian R. Bondy [:bbondy]]

No I looked back at blame history but I couldn't find a reason. 
I'm wondering if we just missed this case originally.

I thought we had this working at some point but I couldn't find a reason within a reasonable time frame.  This is the right fix IMHO though.

Comment 12

[Brian R. Bondy [:bbondy]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/2fd40580507b

Comment 13

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/2fd40580507b