commit 2bba8a47e81d0756a7f4b9718515ff2fd42bde12
author Dominic Farolino <dom@chromium.org> Fri Apr 03 20:46:57 2020
committer Blink WPT Bot <blink-w3c-test-autoroller@chromium.org> Fri Apr 03 21:18:51 2020
tree f51c828040292fec419441ee3d2ba3e5255829c7
parent 7f2d258fafcc211f61eb4da42b798e5167af8f8c

ImageLoader: Stop storing request KURL in Task

This CL removes |ImageLoader::Task::request_url_| in favor of lazily
parsing the image request's absolute URL when ImageLoader::Task is run.

Currently ImageLoader::Task stores a KURL |request_url_|, which is a
snapshot of the image's absolute URL, parsed relative to the image
element's node document at the time ImageLoader::Task is queued.
We shouldn't have to store a snapshot of the parsed URL, because the
source URL at the time an ImageLoader::Task is run is the same as when
the task is queued.

The reason we stored a snapshot of the parsed URL is because what *can*
change in between when an ImageLoader::Task is queued and run, is the
document's base URL, via a `<base>` element. In https://crbug.com/569760
a CL was landed that takes this snapshot, so that if a `<base>` element
was dynamically-inserted before the ImageLoader::Task was run, the URL
would be unaffected by the addition of the `<base>`. However, the spec
is very clear that the parsing of an image request's URL [1] happens
after the microtask is run, meaning our snapshot is unnecessary, and
incorrect.

This CL:
 - Removes the snapshot
 - Fixes the associated test
 - Exports the test to external/wpt so other impls can benefit

[1]: https://html.spec.whatwg.org/multipage/images.html#updating-the-image-data:parse-a-url-2

Bug: 569760,1061685
Change-Id: I8d9e3dbefd27a26626bb930acc809196753c7506
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2103987
Commit-Queue: Dominic Farolino <dom@chromium.org>
Reviewed-by: Kinuko Yasuda <kinuko@chromium.org>
Reviewed-by: Fredrik Söderquist <fs@opera.com>
Cr-Commit-Position: refs/heads/master@{#756376}