[infod-wg] discussion around 'description' field and entity managers
Fisher, SM (Steve)
S.M.Fisher at rl.ac.uk
Wed Apr 5 16:43:48 CDT 2006
Ronny,
You seem to have misunderstood what I meant.
> -----Original Message-----
> From: owner-infod-wg at ggf.org [mailto:owner-infod-wg at ggf.org]
> On Behalf Of Ronny Fehling
> Sent: 31 March 2006 03:21
> To: infod-wg at ggf.org
> Subject: [infod-wg] discussion around 'description' field and
> entity managers
>
> Hi,
>
>
> Steve and I agreed to move our discussion from today's call
> onto an email thread where everybody can participate.
>
>
> Basically, there are two suggestions that came from Steve
> with regards to the interfaces.
>
> 1. To merge
>
>
> BasePublisherManager
> BaseSubscriberManager
> BaseConsumerManager
>
> into a common entity manager. They look a lot alike and have
> essentially the same operations.
I did *not* propose this !!!
> I don't know Steve, what you want to do with the
> BaseSubscriptionManager as it is quite different to the rest...
>
> As Dieter had pointed out a while ago, these Entity-manager
> interfaces could be viewed as a special case of the internal
> INFOD vocabulary; so theoretically, it makes sense to see
> them joined. One counter argument is if it makes the
> interfaces too cumbersome. I actually don't know what the
> original reason was to separate them as I wasn't part of
> INFOD back then. Input?
>
> 2. The second discussion point is a little more sensible.
> Basically, Steve observed that by me adding a description
> field to each interface (as discussed at GGF16), there was an
> explicit vocabulary reference within the entity management interface:
>
> <wsinfod:PublisherDescription Vocabulary="xsd:anyURI">
> { any } ?
> </wsinfod:PublisherDescription> ?
>
> Although intended to be merely a way to include arbitrary
> descriptions (potentially opening up a work-around for
> not-yet implemented functionalities), one could regard this
> from the view of a vocabulary that is referenced this way by
> the entity as an implicit instantiation of that vocabulary.
>
> So, rather than having to create a separate instance of a
> vocabulary and bind it to an entity, we could have the entity
> explicitly reference a vocabulary and with that become
> implicitly an instantiation of that vocabulary.
>
> However, this will have a couple of impacts:
>
> <!--[if !supportLists]-->a. <!--[endif]-->We allow
> multiple vocabulary references for an entity. This means that
> to manage them, alterEntity interface would have to be used -
> is this conceptually correct?
>
> <!--[if !supportLists]-->b. <!--[endif]-->As we allow
> multiple instances of one vocabulary, is the management of
> vocabulary change and the potential impact on the entities a
> harder problem now?
>
>
> c. <!--[endif]-->As we allow multiple associations, is
> the management a harder problem now?
>
> Also, before, the Description field could hold any
> 'blob'-vocabulary and the registry wouldn't care as it
> wouldn't have to parse it. Now, the description vocabulary
> and entry needs to be verified by the registry. We therefore
> loose the back-door for advanced features that we had before.
>
> Also, the Description field used to be an aid to the
> discovery for entities that could facilitate the search for
> specific entities by 'savant' end-users (who could parse the
> description field with its arbitrary vocabulary). We would
> loose that ability - if we don't add a new free-form
> description field.
>
> These are some of the thoughts I had on Steve's comment
> (after finally understanding what he meant in his email ;-) ).
> Any Input / comments?
I think that wsinfod:PublisherDescription should just be xsd:string and
not be associated with any vocabulary. It might say for example
"Purveyor of sausages and fine meats since 1910" or "Widgets made within
24 hours".
The problem I had was the reference "See 3.6 for more details on how to
define a vocabulary for such descriptions." I think this used to be a
string.
If it does have a vocabulary then it is doing the job of the
CreatePropertyVocabularyInstance
Steve
More information about the infod-wg
mailing list