Re: what would be the best way to let users define http ids for bookmarks and topics?

On Wed, 09 Mar 2005 00:27:04 -0500, Eric Prud'hommeaux <eric@w3.org> wrote:

> On Tue, Mar 08, 2005 at 03:50:28PM -0500, Marja Koivunen wrote:
>> Now that the bookmarks basically work, some early thoughts about  
>> letting users definewhether they want to use http ids instead or uuids.
>>
>> The original plan was to also allow users to define a http:// prefix for
>> their bookmark and topic ids when do wishes if they have a domain or let
>> the bookmark server to do that automatically.

When the user as one, it sees a good idea to ask them for their preferred  
location for stuff. Based on the premis that they can then put things  
there. At that point you can buildstuff based on the original names used  
(if they edit the name, you should change the rdfs:label, not the URI, but  
then you need to explain the difference to users between renaming a topic,  
and changing the topics used in a bookmark).

>> This would not create an http file with the bookmark or topic just the
>> id so it does not really matter here if the id is http or not. I did
>> hear discussion stating that the intention should be that the http id
>> files always exists.
>>
>> If I just ask for a prefix, nothing prevents the users to use someone
>> else's domain for the ids, which is not perfect.
>>
>> On the other hand http kind of requires a document to exists and
>> existing http URI does let the user put some info to that. But what  
>> info?

In the case of a bookmark or topic, whatever info you have about the  
bookmark or topic seems obvious.

RDF on the other hand doesn't require anything to exist. It's just "best  
practice" according to some people (including me).

You could also, where they have access to an annotea server, use that to  
store the bookmarks and topics, and so rely on it to tell you the URI.

>> Practically it is better to not include info about the bookmarks or
>> topics in the id file. It is better to use the bookmark file where all
>> other info already is so that we don't have to check all id files in
>> addition to bookmark files.
>>
>> Maybe instead of a file for each URI the user could define a current
>> prefix rdf file and just list all the ids in it? In that case, we need
>> to be careful to not use relative URIs etc. that were discussed earlier
>> in this draft document http://www.w3.org/2003/12/20-local-global.html.
>> Also if it becomes a requirement to add ids to the file when they are
>> created it would again prevent the user creating topics or bookmarks
>> when not online.

Well, if you assume that the file defined by the user is under the user's  
control then they can sync up to it again when they are back online.

This is all a flip side to using other people's definitions, which is just  
a case of reading them. And then you get into the complex area of trying  
to define mechanisms for relating your topics to someone else's -  
extremely useful if you are going to share bookmarks, but somewhat  
complicated.

> This notion involves using attributions and user names. The idea is
> that each user has an area in the filesystem in wich they can create
> whatever hierarchical file system they want. If the user was running
> multiple clients, they would need to do further coordination between
> the clients. For instance, the user tells amaya to use
>   http://iggy.w3.org/annotations/users/marja/amaya/
> and bookzilla
>   http://iggy.w3.org/annotations/users/marja/bookzilla/
> Users could use the names for both the document names and the
> annotation/bookmark(/earl assertion) names.

I am not sure I get what you mean. Can you sketch out a more detailed  
example of what happens please?

cheers

Chaals

-- 
Charles McCathieNevile                      Fundacion Sidar
charles@sidar.org   +61 409 134 136    http://www.sidar.org

Received on Sunday, 20 March 2005 08:58:01 UTC