On 2012-06-15 15:17, Anne van Kesteren wrote:
On Fri, Jun 15, 2012 at 3:11 PM, Julian Reschke <julian.resc...@gmx.de> wrote:
On 2012-06-15 14:52, Anne van Kesteren wrote:
Whitelist.
Sounds like something to avoid.
Yeah it's not exactly pretty, but all implementations have one.
Not sure. How does that handle parsing "http:test" with a base URL
whose scheme is http and host is example.org? Or against no base URL?
You are confusing parsing a URI and resolving it.
They are the same thing in implementations. I don't really see much
reason to separate them.
That is incorrect. Last time I looked at the Mozilla code, it certainly
had a component that parses (not "resolves") URIs without knowing what
happens to them next.
It may be true that it's a single operation when seem from HTML, but
that doesn't necessarily mean that it makes sense to specify it that way.
The RFCs specify how to parse and resolve. I believe the best way to
fill the gap for browser implementations is to specify the error
recovery on top of these operations, instead of pretending the specs are
wrong and rewriting them.
Best regards, Julian
