[Description]: - When the user creates a preference experiment on a previously existing pref and then un-enrolls by disabling the recipe and restarts the browser, the message displayed in the Browser Console is incorrect; It says that the experiment stopped because its value changed, instead of saying that the recipe is not seen anymore. [Affected versions]: - Nightly v66.0a1 [Affected platforms]: - tested on Windows 10, most probably all [Steps to reproduce]: 1. Create a preference experiment recipe on a previously existing pref, approve and publish it. 2. Open the browser with the proper prerequisites and notice that the experiment was run, the preference value was changed. 3. Disable recipe, restart the browser and open the Browser Console. [Expected result]: - Message: " INFO stopping experiment "EXPERIMENT_NAME" because the recipe is not seen DEBUG PreferenceExperiments.stop(EXPERIMENT_NAME, {resetValue: true, reason: recipe-not-seen}) " [Actual result]: - Message: " INFO Stopping experiment "EXPERIMENT_NAME" because its value changed DEBUG PreferenceExperiments.stop(EXPERIMENT_NAME, {resetValue: false, reason: user-preference-changed-sideload}) " [Additional notes]: - The Value of the preference was successfully reverted to its default/original value, despite the message displayed. - The correct message was displayed in the case of disabling a recipe with an experiment conducted on a newly created pref: Message: " DEBUG PreferenceExperiments.stop(EXPERIMENT_NAME, {resetValue: true, reason: recipe-not-seen}) DEBUG PreferenceExperiments.hasObserver(EXPERIMENT_NAME) DEBUG PreferenceExperiments.stopObserver(EXPERIMENT_NAME) WARN Can't revert pref for experiment EXPERIMENT_NAME because it had no default value. Preference will be reset at the next restart. "

Thanks for reporting this. Those log messages come from the lib/PreferenceExperiments.jsm module that I didn't touch, but this still could be my fault. Could you please elaborate on the steps you took to produce this bug? Here's what I tried: - Create an experiment on local dev Normandy modifying a previously-existing pref -- I chose accessibility.typeaheadfind.matchesCountLimit - Enable and publish it - Create a new nightly profile, point it at local dev normandy - Using Normandy dev tools, run Normandy - Verify that the pref is changed - Disable the experiment - Restart Nightly and open browser log -- no log messages yet - Using Normandy dev tools, run Normandy - I saw this message: ``` 1545857019021 app.normandy.preference-experiments DEBUG PreferenceExperiments.stop(ethan-exp-8, {resetValue: true, reason: recipe-not-seen}) 1545857019021 app.normandy.preference-experiments DEBUG PreferenceExperiments.hasObserver(ethan-exp-8) 1545857019022 app.normandy.preference-experiments DEBUG PreferenceExperiments.stopObserver(ethan-exp-8) ``` (The message `stopping experiment "EXPERIMENT_NAME" because the recipe is not seen` doesn't appear in either the Gecko codebase or https://github.com/mozilla/normandy/blob/master/client/actions/preference-experiment/index.js. I'm not sure where that is supposed to be fired.) I also restarted the browser again after I saw this message but no further messages were shown as a result of this experiment.

I cannot reproduce this issue anymore. I am not sure what happened or what I did differently, but I'll catch it again if it will reoccur. I only used this test case: https://testrail.stage.mozaws.net/index.php?/tests/view/842934&group_by=cases:section_id&group_order=asc&group_id=30116