Interoperability of 'The indicated part of the document' for HTML documents

[tkent-google]

Spec: https://html.spec.whatwg.org/multipage/browsing-the-web.html#scroll-to-fragid:the-indicated-part-of-the-document-6
Test: http://w3c-test.org/html/browsers/browsing-the-web/scroll-to-fragid/

In the current specification, a rough algorithm to find "the indicated part of the document" for HTML documents is like:

1. Search IDs for decoded fragid
2. Search names for not-decoded fragid

However, it doesn't match to none of Chrome, Edge, Firefox, and Safari.

Chrome and Safari:

1. Search IDs for not-decoded fragid
2. Search names for not-decoded fragid
3. Search IDs for decoded fragid
4. Search names for decoded fragid

Firefox:

1. Search IDs for decoded fragid
2. Search decoded names for decoded fragid

Edge:

1. Search IDs for decoded fragid
2. Search names for decoded fragid
   (Note that Edge has different behavior in http: and file:.)

I hesitate to make Google Chrome match to the specification because it would introduce yet another incompatible implementation.

[annevk]

Hah, see #2901. It seems that I forgot to test searching <a name> for decoded fragments.

[annevk]

Does Chrome match "top" ASCII-case-insensitively after decoding?

[tkent-google]

Does Chrome match "top" ASCII-case-insensitively after decoding?

Chrome and Safari check 'top' for both of not-decoded fragid and decoded fragid. So #top and #to%70 work. I don't think this behavior makes much sense.

1. Search IDs for not-decoded fragid
2. Search names for not-decoded fragid
3. Check if not-decoded fragid is "top"
4. Search IDs for decoded fragid
5. Search names for decoded fragid
6. Check if decoded fragid is "top"

[annevk]

It seems removing step 3 would not be observable. I don't really care what behavior we end up with as long as everyone aligns.