Re: [Powerbox] New draft based on further collaboration and prototyping

Hi Ricardo,

On Jun 2, 2010, at 13:16 , Ricardo Varela wrote:
> I would like to comment on the concern that until actually the actions
> from "The Prague Doctrine" [1] are completed and we can safely assume
> that we can do a mapping from WebIDL to a RESTful interface (in this
> specific case, to the model Powerbox proposes) I think it may be not a
> good idea to include Powerbox inside the structure of DAP. The reason
> being that the group decided to focus on Javascript APIs and by
> extension on the formal definitions on WebIDL that may be mapped to
> other methods. This is consistent with other initiatives (like JIL or
> BONDI, and now WAC) so also gives us a point of alignment with them.
> If Powerbox is not one of the alternatives that can be used into that
> framework, we would have to basically create specific definitions for
> it, and therefore add fragmentation.
> 
> Moreover, just the fact that developers will have to adequate their
> code for making calls to the APIs by one method (js objects) or
> another (powerbox) can already add extra fragmentation, but that may
> be considered a separate concern, less important/immediate than the
> one i added above (as long as we can prove we can have a wrapper or
> similar for these)
> 
> My opinion is that it may be valid to ask if it belongs at all to the
> items that should be discussed as part of DAP until we have solved
> those issues. Note that I'm not opposing to include this, just to make
> sure first that it doesn't create problems in the long run with the
> expected chartered activity of the group. For example: probably would
> be good to see the experience of implementing a couple of the current
> DAP API proposals in the Powerbox model

The problem is that there's a chicken and egg problem. Either the WG has Powerbox under its umbrella, and there's a point in looking at the issues, or it doesn't and there isn't much motivation to solve them.

If down the line the same APIs are implemented using host objects by half the implementers and Powerbox by the other half, there'll be a small fragmentation problem (small because wrapper code to address it is straightforward). But we can deal with that issue later. My personal expectation is that it's an unlikely outcome, though.

We discussed this to death in Prague (and before) and we came to a consensus that Powerbox was welcome here. Unless there is new technical information being brought in that was unavailable at the time, I won't reopen the discussion. Powerbox is a DAP deliverable.

As also discussed before, it's possible that some APIs won't map well to REST, while others would fit very well. We've agreed that that's fine. For instance, I'm going to stick my neck out there and state that if the following do no map well the REST, then they very likely have design flaws: Calendar, Tasks, Contacts, File System (assuming we keep it), Messaging, Gallery. Capture and System Info may or may not. User Interaction probably won't.

So I hear your concerns but faced with similar arguments made before, the WG has convinced itself that it could survive and deliver nevertheless :)

PS: I'll report on feedback from the JSON Schema community on this pretty soon I think.

--
Robin Berjon
  robineko — hired gun, higher standards
  http://robineko.com/

Received on Monday, 7 June 2010 13:37:20 UTC