Re: [ws-caf] ALS-configuration identifier

[Ricardo Jimenez-Peris <rjimenez@fi.upm.es>]

Mark Little wrote:

> As we've seen   up to  this    point, there  are  different  types  of
> activities,  and   each type is  a    function of  the   ALSs that are
> registered with that specific  activity. So, when creating an activity
> and it's  associated context, there's a requirement   to say "create a
> context for a specific type of  activity". That's essentially what the
> ALS-configuration identifier is,  and with hindsight  I don't know why
> we didn't just explicitly say that begin was  parameterised with the 
> type that'll
> go into the context.
>  
> This can be   accomplished in  a couple of   ways.  One  would be,  as
> described  above, to parameterise   begin.  That assumes that a  given
> context/activity service  can provide  contexts for multiple different
> types   of activity.   The  other  way  would  be  to  have a   single
> context/activity service per activity type (so the paramaterisation is
> essentially   implicit).    Either option   may  make  sense   from an
> implementation perspective,  and we did wonder whether  we  need to define
> this at all.  It could be an  addressing issue: does a  service manage
> multiple "sessions" or only one - that could be  hidden by the service
> implementation. What we're saying is that how you, as the creator of 
> an activity,
> locate the right activity service/context service is always going to 
> be outside the scope
> of the specification in one way or another.

A parameterised begin would be a nice thing to have, since it would not 
force implementations
to take the second option. The second option can always be taken anyway.

Ricardo

-- 
Ricardo Jimenez-Peris
Associate Professor
Universidad Politecnica de Madrid
http://lsd.ls.fi.upm.es/%7Erjimenez/