Re: testing URL decomposition (related to ISSUE-56)

On Feb 11, 2010, at 12:22 AM, Julian Reschke wrote:

> Aryeh Gregor wrote:
>> On Wed, Feb 10, 2010 at 8:49 AM, Julian Reschke <julian.reschke@gmx.de 
>> > wrote:
>>> I don't see any kind of interoperability here, even not for the  
>>> simplest
>>> test cases.
>> Real-world usage is often simpler than even the simplest test-cases.
>> Typical URLs contain no spaces, no Unicode characters, and no weird
>> stuff like multiple hash characters.  These APIs are realistically
>> useful even if they start to become unreliable for slightly less
>> common types of URLs.
>
> Even if they don't agree whether the path starts with "/" or not?
>
>>> Maybe it's time to deprecate this mess (does anybody use this?)
>> Yes, definitely.  I can find a whole bunch of uses in MediaWiki, for
>> instance.  The existing interface will have to be specced anyway.
>
> The question is how to spec it. Right now the UAs vary *a lot*. I  
> recommend that UA implementers have a look at the test results and  
> provide feedback whether they want to change their implementations.
>

Of the browsers I tried (Firefox, Opera, Safari, Chrome), Firefox had  
the most green and I don't see any problem with changing Safari to  
match.

There is one behavior checked by these tests that seem potentially  
unwise: trying to differentiate between URLs that had particular  
percent-escaping in the original source vs. equivalent ones that did  
not. All browsers I tried seem to either consistently escape or  
consistently unescape in any given component. Unless IE differentiates  
I don't think we can conclude that this is required behavior.

Regards,
Maciej

Received on Thursday, 11 February 2010 09:07:14 UTC