[SysInfo] Status of AI to review

Hi all,
I am working on the action item I have to provide input on the SysInfo
API, and will upload detailed comments soon. Here are some general
comments based upon what I have done so far.

In general, I echo the comments of Jose on the UWA list that the overlap
with the DCO is a concern, in that we need to ensure that the DAP API
does not veer off from the ontology/vocabulary approach established in
the DCO. There are new properties and ways to express existing
properties being proposed, and it's very important that there be
alignment. The UWA has been also working to further ensure
convergence/synergy with the existing standards defined by the OMA
(UAProf and DPE), which are by far the most broadly supported device
property vocabularies, and in active maintenance and evolution within
the OMA. Being chair of the OMA CD group and WI champion for the Device
Capability and Mobile Client Environment (browsing) work items in OMA,
it's a priority of mine (and I'm sure, other OMA members) to ensure that
DAP does not complicate that ongoing cross-SDO alignment. The SysInfo
API can in fact be a way to strengthen the alignment and cross-SDO
cooperation, if we move to define it carefully and consistently.

For that purpose I am preparing a detailed mapping between the property
vocabularies of UAProf/DPE, DCO, existing DOM navigator objects, and DAP
SysInfo as proposed. This should be ready for review and discussion
before the Prague meeting.

This analysis will be usable as a guide to ensure alignment on the
properties that are in common, and facilitate discussion on the priority
of those that are new. It's my opinion that the most important thing for
the DAP SysInfo API is to establish a *pattern* for how properties are
exposed, and to keep that pattern consistent as possible. The property
set is an important but secondary concern. Undue focus on the property
set at this point will likely lead to vocabulary fragmentation and
specification of an inconsistent set of interfaces for the various
property types. I already see this developing in the SysInfo API draft,
and that's why I still recommend that we use a much simpler approach
closer aligned with the BONDI devicestatus API.

Thanks, 
Bryan Sullivan | AT&T

Received on Wednesday, 17 February 2010 07:31:07 UTC