- From: Simon Sapin <simon.sapin@kozea.fr>
- Date: Fri, 18 May 2012 00:11:43 +0200
- To: www-style@w3.org
Le 17/05/2012 17:14, Tab Atkins Jr. a écrit :
> I know the RFC we currently point to doesn't fit that definition.
> I'm hoping the URL spec from abarth will.
Which spec is that?
> #2 is great. We just need an "always-invalid URL" to resolve it to,
> like"about:invalid" or something.
As noted in another message ("Other test case: url('%é') is turned to
%%C3%A9" ...) apparently, the browsers do not care at all about URL
validity. They accept any Unicode string and do something like in the
IRI RFC to encode the string into an URI.
http://tools.ietf.org/html/rfc3987#section-3.1
Only then, fetching the URI may fail for various reasons (HTTP 400 Bad
Request in this example) with the same effect as "about:invalid" (except
for the cost of a network round-trip.)
--
Simon Sapin
Received on Thursday, 17 May 2012 22:12:14 UTC