Steps to reproduce
The test case should say “everything is okay.”
Dev Channel specific:
Comments and activity
- Microsoft Edge Team
Changed Assigned To to “Tony S.”
Changed Assigned To to “Venkat K.”
Changed Assigned To from “Venkat K.” to “Shawn P.”
Changed Status to “Confirmed”
Changed Status from “Confirmed” to “Won’t fix”
Changed Assigned To to “Ali A.”
Changed Status from “Won’t fix”
I have another test case demonstrating IDB getting into a state where it stops responding: http://bl.ocks.org/nolanlawson/raw/2735f83832f8d03e4b25d53c41fed702/
This repro is based on the PouchDB performance test suite, which deletes databases and recreates databases while querying and applying upgrades. Typically you can succeed in getting Edge/IE to stop responding after only loading the page once. You can also play with the query parameters to change the number of iterations that the test is run (I have it set to 5, but even 1 is enough to repro if you reload the page enough times).
To work around this issue, I often will just clear all browser data or modify my hosts file so that I can use a new origin. It seems to be a race condition, because it cannot be reproduced consistently.
Note that Chrome and Firefox handle this page just fine, even if you reload while the test is in progress or run the test several times.
I can repro in Edge using rs_onecore_webplat 5021.1000.
We have experienced the same thing. Actually we added a browser test page to make sure people had proper browsers with functioning features, which is where we discovered this. I will provide a link in a minute with our example of how we cause the issue - which is straight native methods in JS - no library involved at all.