Re: Draft Charter for a Device API and Security WG

Hi Art, all,

here are comments on your charter proposal on behalf of Vodafone.

On May 11, 2009, at 12:37 , Arthur Barstow wrote:
> Attached is a new Draft charter for a single WG to define both the  
> Device Service APIs and Security Policy work we've been discussing  
> (as a result of the December 2008 Workshop) [1].

By and large we support this charter, and fully intend to commit  
resources to this WG once it has been ratified. We would like to urge  
the team to move with haste on the creation of this WG. We are also  
glad to note that the WG is intended to operate in public.

Editorial: I wouldn't include "All of the API specifications will use  
Web IDL to describe the API." I agree with the statement but it's too  
restrictive to be in a charter.

Substantial: We are not wild about the "Device API Design Patterns".  
The timeline puts it on Rec track which we doubt is useful. We agree  
that making the APIs consistent amongst themselves is desirable, and  
that documenting the decisions we made is useful for the community at  
large, but there should be no normative dependency on such a document  
and it shouldn't be on the critical path — we've done APIs before  
without it. We therefore propose that this document be developed on  
the side (i.e. not before the others) and be a Note, not a Rec.

Editorial: Do the PIM APIs need to be grouped? It appears that they  
could probably just be used separately, like any of the other APIs.

Editorial: We would like it to be clear that the WG may produce  
multiple versions of an API, and that the plan is to push out  
consensual, technically simple APIs fast, and build on top of those to  
add features that may take more time to define and reach agreement on.

Substantial: Do we really want to have test suites ready for the end  
of LC? The usual is CR. We don't necessarily object, but the effect  
might simply be to lengthen LC and shorten CR.

Substantial: While we realise that charters are expected to contain  
dates picked out of a hat in a table called "Milestones", and that  
documents will subsequently be released on any date except those  
chartered, we do have some concerns about the underlying assumptions  
that may have guided the this particular "Milestones" table. First and  
foremost it seems to put the DAPIDP document before the rest, as if a  
primacy and dependency were intended — as outlined above we do not  
believe that this would constitute the best use of the WG's time.  
Another concern is that the API specifications appear to move in  
perfect unison, which is both unrealistic and undesirable. We propose  
that this table be removed, and replaced with text indicating that: 1)  
DAPIDP is expected to be developed as needed on a voluntary basis and  
may be (re)published as a Note at irregular intervals, 2) that all  
documents will have reached Rec before the end of the charter, and 3)  
that otherwise the WG will control the timing of it releases, and may  
for instance identify a subset of its work as high-priority and decide  
to fast-track it. It is our belief that this better reflects what will  
actually happen irrespective of what the charter looks like — so we  
might as well be honest up front.

Editorial: In the interest of helping our community find us, we  
suggest that the group's page not be in a dated URI.

Other than that, it's all good.

-- 
Robin Berjon - http://berjon.com/
     Feel like hiring me? Go to http://robineko.com/

Received on Tuesday, 12 May 2009 15:08:05 UTC