1753917 - FxMS private browing integration test fails with `--verify` in TelemetryTestUtils.jsm

Status: RESOLVED DUPLICATE of bug 1754546

(Firefox :: Messaging System, defect, P1)

Description

[Dan Mosedale (:dmosedale, :dmose)]

Running

./mach test --headless --verify browser/components/privatebrowsing/test/browser/browser_privatebrowsing_about_nimbus.js

Results in the following failures on current mozilla-central, which does not yet

Doing it without --verify seems to go fine.

browser/components/privatebrowsing/test/browser/browser_privatebrowsing_about_nimbus.js
  FAIL parent must be in snapshot. Has []. - false == true - JS frame :: resource://testing-common/TelemetryTestUtils.jsm :: assertEvents :: line 143
Stack trace:
resource://testing-common/TelemetryTestUtils.jsm:assertEvents:143
chrome://mochitests/content/browser/browser/components/privatebrowsing/test/browser/browser_privatebrowsing_about_nimbus.js:test_experiment_messaging_system:545
  FAIL Uncaught exception - at resource://testing-common/TelemetryTestUtils.jsm:167 - TypeError: can't access property "map", snapshot is undefined
Stack trace:
assertEvents@resource://testing-common/TelemetryTestUtils.jsm:167:20
test_experiment_messaging_system@chrome://mochitests/content/browser/browser/components/privatebrowsing/test/browser/browser_privatebrowsing_about_nimbus.js:545:22
  FAIL Test timed out -
  FAIL Found a browser window after previous test timed out -

I did a quick look through the code and up the stack, and it looked to me like the issue was in the C++ telemetry code returning NS_ERROR_FAILURES that didn't have documentation about how one should handle those failures. It's entirely possible that I just missed something on my quick skim, however.

Comment 1

[Dan Mosedale (:dmosedale, :dmose)]

One the one hand, this is "just" with --verify, on the other, it makes --verify less useful with this particular file at least, since one can never verify that everything works because of this issue.

Comment 2

[Dan Mosedale (:dmosedale, :dmose)]

Note that this happens on the trunk, where bug 1749775 (private browser telemetry for the dismiss button) has not yet landed.

Comment 3

[Dan Mosedale (:dmosedale, :dmose)]

This will also certainly end up being a DUP of bug 1754546.