Re: parsing URI (references) according to RFC 3986

On 2011-06-20 10:47, Adam Barth wrote:
> On Mon, Jun 20, 2011 at 1:13 AM, Julian Reschke<julian.reschke@gmx.de>  wrote:
>> On 2011-06-20 10:03, Adam Barth wrote:
>>> Even just trivial things need to be cleaned up, like:
>>>
>>> http://ExAmple.CoM/
>>
>> What needs to be cleaned up here?
>
> * FF canonicalize('http://GoOgLe.CoM/') is 'http://google.com/'
> * IE canonicalize('http://GoOgLe.CoM/') is 'http://google.com/'
> * KR canonicalize('http://GoOgLe.CoM/') is 'http://google.com/'
> * SA canonicalize('http://GoOgLe.CoM/') should be http://google.com/.
> Was http://GoOgLe.CoM/.
>
> IE, Firefox, and Chrome convert host names to lower case.  Safari does not.

Yes. Why is this a problem? What does it have to do with URI/IRI parsing 
or resolution?

>>> http://www.example.com/##asdf
>>
>> Either reject the reference as invalid, or treat this as a fragment with
>> value "#asdf".
>>
>> *How* to handle fragments depends on media types, not URI parsing, so I'm
>> not sure we should try to answer this here...
>
> FF canonicalize('http://www.example.com/##asdf') is
> 'http://www.example.com/##asdf'
> IE canonicalize('http://www.example.com/##asdf') is
> 'http://www.example.com/##asdf'
> KR canonicalize('http://www.example.com/##asdf') is
> 'http://www.example.com/##asdf'
> SA canonicalize('http://www.example.com/##asdf') should be
> http://www.example.com/##asdf. Was http://www.example.com/#%23asdf.
>
> The question is whether # occurring in the fragment should be coerced
> to be %-escaped.  My reading of the evidence here says "no."

Again, what's to do here?

You can observe this in the DOM, so it's a DOM issue.

You can observe the *behavior* when navigating, and that's a media type 
issue.

> On Mon, Jun 20, 2011 at 1:34 AM, Julian Reschke<julian.reschke@gmx.de>  wrote:
>> On 2011-06-20 10:24, Chris Weber wrote:
>>> 6) Handling percent-encoded values in various components
>>
>> Is there a *problem* related to this?
>>
>> I can see that the exposed DOM properties vary on how things are
>> canonicalized, but that's a DOM issue, not a URI/IRI issue.
>
> You can play games about who needs to spec this stuff, but it needs to
> be specced.  In implementations, this work is done by the URL
> processing code, not by the DOM processing code.

I'm not playing games, I'm trying to understand what needs to be done 
*here*. And I don't believe this belongs here (nor other DOM questions 
like how the port number defaulting works).

Best regards, Julian

Received on Monday, 20 June 2011 08:58:15 UTC