[locators] Re: While it is still fresh in our minds: '!' is not just a funny fragment identifier...

(I have added a [locators] prefix to the subject. It may be a good practice for those who use such prefixes for mail filtering…)

Hi Romain,

> On 21 Dec 2015, at 19:17, Romain Deltour <rdeltour@gmail.com> wrote:
> 
> 
>> This is a major difference that we should not forget about.
> 
> Absolutely, right.
> 
> I was more thinking in terms of spec work:  we should not try to (re)invent the wheel and touch fragment IDs where they're already well-defined (like HTML), but on the other hand, for new media types (for instance a JSON PWP manifest?) we have new grounds to explore and it may be relevant to consider at a fragment identifier-based approach (which is, as you correctly point out, technically different from a custom-URL-separator-based approach).
> 

We agree! (And I clearly misunderstood a remark on the call.) And I indeed think any discussion on a possible fragment ID is premature at this point although it may come up later if we need it.

A side issue: in theory, registering a fragment ID is relatively easy if part of the definition of a new media type. But it seems to be complicated, and I am not even sure how this should be done, if it is meant for an existing data type. Ie, where is such a definition should be registered, documented, etc. (Heather might know something on this, though…) The XML world has defined the xpointer scheme which allows for the registration of XXXX#foo(bar) type fragment identifiers (cfi has been registered that way I believe) but that mechanism has been defined for XML syntaxes only. I do not know how one would, say, define and register, formally, a new fragment ID for HTML. One more reasons to try to keep away from all this if we can…

Ivan


> Romain.
> 
>> On 21 Dec 2015, at 18:21, Ivan Herman <ivan@w3.org <mailto:ivan@w3.org>> wrote:
>> 
>> This came up today, I think maybe Romain mentioned it: that the '!' approach for content URL looks very much like a fragment ID, so why do we make a differentiation? (But I may have misunderstood the remark, in which case my apologies!)
>> 
>> There is one aspect that we should not forget about where '!' and '#' are very different. Per HTTP the fragment identifier is resolved, and acted upon, on the client side. Ie, the approach is that if I request
>> 
>> http://www.example.org/A#B <http://www.example.org/A#B>
>> 
>> then the GET request will deliver the http://www.example.org/A <http://www.example.org/A> as a whole to the client, which will then select, in a second step, B out of A.
>> 
>> However, a '!' is a bona fide part of a URI. Ie, if I request
>> 
>> http://www.example.org/A!B <http://www.example.org/A!B>
>> 
>> then the server is supposed to deliver http://www.example.org/A!B <http://www.example.org/A!B> to the client, not http://www.example.org/A <http://www.example.org/A> (whatever that is).
>> 
>> This is a major difference that we should not forget about.
>> 
>> Happy holidays and lots of rest to all of you/us!
>> 
>> Ivan
>> 
>> 
>> 
>> ----
>> Ivan Herman, W3C
>> Digital Publishing Lead
>> Home: http://www.w3.org/People/Ivan/ <http://www.w3.org/People/Ivan/>
>> mobile: +31-641044153
>> ORCID ID: http://orcid.org/0000-0003-0782-2704 <http://orcid.org/0000-0003-0782-2704>
>> 
>> 
>> 
>> 
> 


----
Ivan Herman, W3C
Digital Publishing Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
ORCID ID: http://orcid.org/0000-0003-0782-2704

Received on Tuesday, 22 December 2015 08:14:35 UTC