Re: Orphaned annotations

On Mon, 18 Mar 2002, Jose Kahan wrote:

> Thanks for your remarks concerning the Annotea page. Yes, we could have
> added ID pointers, but this slipped by. On the other hand, note that
> all the news items on the W3C home page do have a distinct ID attribute
> (searc in the source code for <div class="item id="something">. I also
> remember some talk about producing all of our specs with ID attributes.

ID attributes for everything are fine for a scenario where all contents
are tightly controlled by a publishing system that manages them.
But that's a rather restricted scenario, and clearly not applicable
to the Web at large.

> We have never tried to hide this limitation of XPointer. You call them
> fuzzy pointers, but in reality, it's a limitation of XPointer. We have
> identified two problems for XPointers, in both our paper and in the Amaya
> on-line documentation:

Jim and I worked through several implementations of fuzzy pointers
in the course of interfacing our respective software agents.  What
we have now appears to be equivalent to Annotea, except in that it
identifies an element and not a range.

They are XPointers into a normalisation of the document markup -
hence fuzzy (as Jim calls them).

> - orphan annotations: an annotation can't be attached anymore to a document.
> - misleading annotations (this is what you refer to as fuzzy pointers).
>   An annotation is attached to a wrong place in the document.

No it isn't!  Using fuzzy pointers, Jim's client software can query
my server, and does NOT misplace an earl assertion (or annotation).
The problem isn't fuzziness, it's change!


> Besides promoting the use of the ID attribute, we don't have a any good
> solution today to make XPointer more robust.

I'm making one such suggestion!

 There are some ideas floating
> around to solve this problem, some of them proposed by the Annotation group
> at Microsoft. The consensus seems to be that it's better to have orphan
> annotations than a possibility of misleading annotations.

But we have both!

-- 
Nick Kew

Site Valet - the mark of Quality on the Web.
<URL:http://valet.webthing.com/>

Received on Monday, 18 March 2002 15:27:58 UTC