Conversation
|
Rather than a random key, which makes the test results a bit harder to reproduce - could we instead clear all the cookies at the end of the test and each test can use a it's own key per test (e.g. |
|
@f2prateek agreed for the reproducibility. One note: this will make the tests flaky when run concurrently, cookies are shared across browser instances so it could technically create a race condition (which I did get pretty often when stressing karma to try to trigger the flaky tests) :
Removing all cookies before/after each test worsen the problem. I don't think we're affected by this on our CI environments (IIRC SauceLabs isolates each instance) but wanted to point it out. Pushed a commit with deterministic tests 👍 |
Codecov Report
@@ Coverage Diff @@
## master #91 +/- ##
=======================================
Coverage 98.58% 98.58%
=======================================
Files 11 11
Lines 636 636
=======================================
Hits 627 627
Misses 9 9Continue to review full report at Codecov.
|
On a total of 28 tests :
cc @f2prateek
Ref: https://segment.atlassian.net/browse/LIB-354