I think simonp@opera.com wrote this test for Presto originally.

HTMLMediaElement::parseAttribute actually doesn't do anything when removing the src attribute, but it looks like this test is failing due to another bug.

The test has two scripts, and that is because per spec, the resource selection algorithm should "await a stable state", which is now equivalent to queuing a microtask. So per spec, when the second scripts starts running, the resource selection algorithm should already have looked at the src attribute and done nothing because its value was empty.

At that point, removing the src attribute shouldn't do anything additionally, and if it did kick of resource selection the error event would fire on the source element already added.

This doesn't work in Blink because the resource selection algorithm doesn't queue a microtask to continue, but instead starts a timer, which is more like a normal task, so it won't have run by the time the next script is encountered.

This is also why several others of these tests fail, at least some of those that fail on networkState being 0 while expecting 3 or vice versa.

I'll file a new tracking bug and update https://docs.google.com/spreadsheets/d/1-xlh6co6QYC-KVof27sVxjA7yuUFvYcaul3vE5dGHso/edit#gid=0

I've filed issue 593289 and linked an number of test failures to that in the spreadsheet. tkent@, if you like you can dupe this issue into that one, but it doesn't really hurt to track them separately either.

Renaming Blink>Audio to Blink>Media>Audio for better characterization

Renaming Blink>Video to Blink>Media>Video for better characterization

Some compat info:
- Works on Safari and Edge;
- Broken on Chrome and Firefox.