Re: HTTP URI scheme

Julian Reschke wrote:
> Werner Donné schrieb:
>>
>> Hi,
>>
>> Implementing an efficient client is a challenge if you have to
>> discover the capabilities at run-time. Either you end up with
> 
> Where exactly would a "webdav" scheme help here? Can you give a 
> definition what information it would provide? Would I be allowed to use 
> it for resources that just support PROPFIND, for instance?

No, of course not. The scheme tells you you can rely on all the
mandatory features of the WebDAV protocol, which allows you to
switch to another protocol handler without the extra round-trip
to find that out.

A failure in that case can be considered as a low-level system failure,
which is in general handled with much simpler logic than what you
need if you have to be very lenient in the middle of application
logic. The former is an exception and can be dealth with as such.

> 
> -> Granularity problem, for once.
> 
>> a lot of OPTIONS method calls or a lot of state to avoid them.
>> Another alternative is recovery code for failures all over the
>> place, because you can't just fail without telling the user
>> in a proper way what the problem is, which depends on the
>> scenario he is in. "Not implemented" is not good enough when
> 
> That's why the newer WebDAV specs provide DAV:error. And again, how 
> would a different URI scheme help here?

The DAV:error just gives you a report about the point of failure.
It is not necessarily the case that the action of the user is
directly related to that point. He could have launched a procedure
which comprise much more than what you would typically expect
behind a link in a browser.

> 
>> it is about something deep inside some logic that the user has
>> triggered, because he will not be able to related it to what
>> he did.
>>
>> Why exactly what Apple did was a mistake?
> 
> For once, because they forced new identifiers on the web, without taking 
> care of specifying what they mean (or did I miss a scheme registration?).
> 
> As far as I can tell, the only way they differ from HTTP URLs is that 
> Apple's software uses them to invoke a different user agent. You don't 
> need an URI scheme for that, a MIME type works just fine (see, for 
> instance, <http://greenbytes.de/tech/webdav/rfc4709.html#rfc.section.A.2>).
> 
> Best regards, Julian
> 

-- 
Werner Donné  --  Re
Engelbeekstraat 8
B-3300 Tienen
tel: (+32) 486 425803	e-mail: werner.donne@re.be

Received on Friday, 12 January 2007 10:25:42 UTC