Clarify RT entry creation when iframe changes src

[noamr]

Take the following scenario:

• Load an iframe with a src
• When the iframe loads, the owner document changes the src to a different URL.

In WebKit/Chromium, there would be 1 resource entry in the performance timeline. In Gecko there would be 2.
The previous Resource Timing spec was a bit vague about it, where the scenario presented there was about internal iframe navigations.

I believe that WebKit/Chromium try to follow the spirit of the old spec, but that Gecko is doing the actual right thing (there's no point in hiding the resource entry for a resource loaded explicitly by the owner document).

This affects the iframe bit of whatwg/html#6542.

[noamr]

See WPT: https://github.com/web-platform-tests/wpt/pull/32490/files#diff-cd501c053b40769b623223a45356d4b2468dcbdc326a6c5c5c17758d1f832b6eR26

The second test succeeds in Firefox and fails in Chrome/Safari.

[yoavweiss]

That makes sense. It's possible that Chromium's implementation of the restriction was over-zealous...

/cc @achristensen07

[achristensen07]

Acknowledged. I don't have a strong preference either way

[yoavweiss]

@noamr - can you file a crbug on this?

[noamr]

@noamr - can you file a crbug on this?

I was going to once I file the HTML PR, as right now this is unspecified. But I can start with a crbug.

[noamr]

@noamr - can you file a crbug on this?

Done. https://bugs.chromium.org/p/chromium/issues/detail?id=1290721
and webkit https://bugs.webkit.org/show_bug.cgi?id=235561

together with the spec PR https://github.com/whatwg/html/pull/7531/files

[noamr]

whatwg/html#7531 and the new WPTS clear this up.