On 2011-10-11 00:30, Ian Hickson wrote:
On Sun, 9 Oct 2011, Arthur Barstow wrote:
On 10/7/11 8:32 AM, ext Julian Reschke wrote:
As far as I recall, we agreed in the IETF WG that parsing of web socket URIs
should work exactly the same way as for any other URI scheme. It appears
that the API spec now tries to override this, and this looks problematic to
me.
On 10/7/11 9:30 AM, ext Arthur Barstow wrote:
In [1], Julian asks about Web Socket API rev 1.247 [2], the change that adds
the Parsing WebSocket URLs section (CVS comment "Revert the part of r5409 that
removed the URL parsing algorithms, since it's no longer defined in the
protocol spec. (whatwg r6632)").
Would you please elaborate on this change?
On 10/7/11 11:28 AM, ext Ian Hickson wrote:
Elaborate in what way?
Why is the override in 1.247 needed, given what Julian indicates above?
There's no override. It's just defining how you do it because nothing else
defines it.
As far as I can tell, it's a mix of things repeated from
<https://tools.ietf.org/html/draft-ietf-hybi-thewebsocketprotocol-17#section-3>,
things that may be useful, and things that do not make sense at all.
In particular:
"Resolve the url string, with the URL character encoding set to UTF-8.
[RFC3629]"
"Resolve" is undefined as far as I can tell, in particular it's not
clear at all what "URL character encoding" means in this context.
Best regards, Julian
