On Sat, Jul 18, 2020 at 4:09 PM Larry Masinter <LMM@xxxxxxx> wrote:
I didn't mean to suggest that merely by publishing the URL standard that it would automatically change IETF standards.
(But if you had to choose an implementation, one of which is compatible with the browsers, and one which is compatible with the RFCs, which would you choose?)
I would probably use the WHATWG version where applicable, but it doesn't cover enough to obsolete RFC 3986 and RFC 3987, in particular the appendix on "Delimiting a URI in Context" [1].
That appendix is becoming obsolete, whether we like it or not, as users have come to expect behavior similar to what Twitter does. See [2] for an example of a URI delimited in an unexpected way due to a user making this assumption.
I wrote it all down in [3]. That code also covers normalization and IDNA outside of the grammar (mostly for length calculations). The code contains a few Twitter-specific things (like t.co processing and some strange rules around parentheses in a path), but it is otherwise pretty general. In particular, it covers what they call a "URL without protocol", which I'm sure will annoy some.
thanks,
Rob
