- From: Carsten Bormann <cabo@tzi.org>
- Date: Tue, 14 Jun 2022 13:23:51 +0200
- To: Daniel Stenberg <daniel@haxx.se>
- Cc: Roberto Polli <robipolli@gmail.com>, HTTP Working Group <ietf-http-wg@w3.org>, Giuseppe De Marco <giuseppe.demarco@teamdigitale.governo.it>
On 14. Jun 2022, at 12:25, Daniel Stenberg <daniel@haxx.se> wrote: > >> I noted that some specs reference URL/URI from https://url.spec.whatwg.org/ >> while others use https://www.rfc-editor.org/rfc/rfc3986 & co. > >> I assume that those specs are aligned, isn't it? > > They are very much not aligned. There are plenty of ways we can write URLs and URIs that parse differently based on these specs. I collect some of those issues here: https://github.com/bagder/docs/blob/master/URL-interop.md WHATWG describes Browser URLs. URIs are used outside browsers, and even with different schemes than http/https, so it is good that we have a reference for URIs (3986) and IRIs (3987) in general. The RFCs are also a stable reference, which is useful for specifications that don’t want to (or can’t) track WHATWG. I think that 3986/3987 would be useful subjects for a revision, but I’m not sure there is energy for that. On a related note, we have started extracting the data model from the URI syntax and written it up in https://www.ietf.org/archive/id/draft-ietf-core-href-10.html — that was an interesting exercise (the idea is to replace the usually abysmal URI parsers found in low-resource IoT space by converting URIs/IRIs into CRIs before they enter the constrained domain; this is essentially a slight generalization of what CoAP already does). Grüße, Carsten
Received on Tuesday, 14 June 2022 11:24:10 UTC