1597669 - Users are wrongly enrolled in studies with attribution code after uninstalling the "custom" build and installing a regular one

Status: RESOLVED INVALID

(Firefox :: Normandy Client, defect)

Description

[Ciprian Muresan [:cmuresan], Formerly of Ecosystem QA - please needinfo @mcoman for things regarding the team]

[Affected Platforms]:

• All Windows

[Affected Versions]:

• Firefox Nightly 72.0a1, Build ID 20191119043902
• Firefox Beta 71.0b10, Build ID 20191114160003

[Prerequisites]:

• Have a stub installer with attribution code downloaded on your computer.
• Have a regular stub installer downloaded on your computer.
• Have a recipe that targets users that have used the stub installer with attribution code.
• Firefox is not installed.

[Steps to reproduce]:

1. Install Firefox using the stub installer with attribution code from prerequisites.
2. Navigate to about:studies and observe the page.
3. Uninstall Firefox.
4. Install Firefox using the regular stub installer from prerequisites.
5. Navigate to about:studies and observe the page.

[Expected result]:

• Step 2: The experiment from prerequisites is displayed in the list.
• Step 5: No experiment is displayed in the list.

[Actual result]:

• Step 2 & 5: The experiment from prerequisites is displayed in the list.

[Notes]:

• The issue is not reproducible if the regular stub installer is installed first, meaning that no experiments are displayed after installing.
• It seems like the regular stub installer does not clear the previously set attribution code.

@mythmon, @rhelmer, could you please take a look at this?

Comment 1

[Robert Helmer [:rhelmer]]

I think this is expected behavior, that a vanilla pave-over install will not clear existing postSigningData file, and therefore will not override existing attribution data.

Molly, does this sound right to you? I'm not sure what expected behavior is wrt pave-over installs.

Comment 2

[Molly Howell (she/her) (mostly inactive)]

Yes, it is expected behavior; we only clear the existing attribution if the installer has some new data to overwrite it with, on the theory (I believe) that if the new installer came from somewhere that doesn't have its own attribution, that means the previous attribution is still current and correct.

Comment 3

[BugBot [:suhaib / :marco]]

The priority flag is not set for this bug.
:mythmon, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 4

[Ciprian Muresan [:cmuresan], Formerly of Ecosystem QA - please needinfo @mcoman for things regarding the team]

Marking this issue as RESOLVED-INVALID based on the explanations given by Molly and Robert.