[Standards] geoloc: PEP=user, <iq>=resource

Adam Nemeth aadaam at gmail.com
Thu May 31 16:56:48 UTC 2007


Hi all,

although answering only this letter, the answer is for the thread...

On 5/29/07, Ian Paterson <ian.paterson at clientside.co.uk> wrote:
> Adam Nemeth wrote:
> > I still think Geoloc is meaningless:
>
> No, resource is meaningless (i.e. "opaque") ;-)
>

Probably just the name is bad: For geoloc, I mean: geographical
information, that is exactly translatable to a GPS-coordinate set
(that's why it's called geographical.) While it's not a bad
information to know sometimes, it does not specify my 'virtual clone'
as much as it could be necessary - when I have a bunch of presence
enabled devices, clients, websites, games, etc at one - geographical -
place.

So Geoloc is limited by its name itself: it contains the user's
geographical information, nothing else. If it does contain something
else, however, then it could be renamed to something else, or a
different XEP could emerge, for example, Resource Descriptor. So
basically we have a kind of new information, which isn't really a
presence nor a message:

While presence and geoloc informations are eventing-like - they could
change in certain situations in a session's lifetime - resource
informations are session-constant, just like discovery items - your
car won't turn to be something else, or if it does, you have seen too
much Transformers I think ;)

So I propose - without any intentions against geoloc as an eventing /
presence protocol itself- to make a new XEP about resource identity,
or better:

Make it part of XEP-0030, and encourage implementations to use this
information instead of the original resource string, if found.

Example:

Use romeo at montague.net/orchard when you get:


<iq type='get'
    from='juliet at capulet.com/21323131233423423'
    to='romeo at montague.net/324324234234'
    id='info4'>
  <query xmlns='http://jabber.org/protocol/disco#info'/>
</iq>

<iq type='result'
    from='romeo at montague.net/312211223'
    to='juliet at capulet.com/342234234234'
    id='info4'>
  <query xmlns='http://jabber.org/protocol/disco#info'>

    <identity category="resource" type="human" name="orchard"/>

    <identity
        category='client'
        type='pc'
        name='Gabber'/>
    <feature var='jabber:iq:time'/>
    <feature var='jabber:iq:version'/>
  </query>
</iq>

(I didn't want to overwrite client identity - so you know when clients
use Gaim / Psi / Adium etc - nor introduce another attribute for
identity - like description - even if it would be probably more
accurate for most cases)

What do you think?

(Sorry for the list if this mail gets here twice, hope not.)

> >     1) I don't want to fill in a form, since "office" means just what I
> > want to mean (see Dave's scenario)
>
> XEP-0080 doesn't say the user has to fill in a form. The client UI is
> implementation specific. A client could employ the geoloc <description/>
> field with exactly the same user interface you are using today for your
> resource identifier.
>
> >     2) A desktop computer usually does not have a GPS
>
> Geoloc can be limited to just the <description/> field.
>
> >     3) A mobile client could be used even inside a house, when location
> > is the same, resource is different....
>
> The 'type' (and 'name') attributes of the disco <identity/> element
> enable this to be communicated.
>
>
> The typical things clients are doing with overloaded resource identifier
> strings can all be achieved in a *standard* way with the geoloc
> <description/> field and/or the disco <identity/> element.
>
> - Ian
>
>


-- 
Aadaam <aadaam at gmail.com>



More information about the Standards mailing list