- From: Arthur Barstow <Art.Barstow@nokia.com>
- Date: Mon, 2 Feb 2009 06:59:37 -0500
- To: ext Marcos Caceres <marcosscaceres@gmail.com>
- Cc: public-webapps <public-webapps@w3.org>
Marcos, On Jan 31, 2009, at 9:43 AM, ext Marcos Caceres wrote: > Hi Art, > On Sat, Jan 31, 2009 at 2:18 PM, Arthur Barstow > <Art.Barstow@nokia.com> wrote: >> Does the P&C spec include any normative assertions that create a >> dependency >> on the API & Events spec? I just made a quick scan of the P&C's >> latest ED >> and I did not find any any such statements. >> >> If there are no such dependencies - and my expectation is there >> should not >> be - then, SHOULD in the second sentence above is the strongest >> statement >> that can be made. Any of: using MAY, making the statement as non- >> normative >> text, or removing it altogether would be appropriate if there is >> no such a >> dependency. >> >> If there is such a dependency, then that's probably a bug we need to >> discuss. > > Ok, I see your point. We do not have any dependencies that I can > think of. > > However, please allow me to explain my rationale for that section as I > still feel it is important. The Widget User Agent section of the P&C > spec defines what a widget specs and other specs (e.g., gif, jpg, etc) > need to be brought together to create a "widget user agent". So, in a > sense, this section is unrelated to the P&C spec; but it needs to be > stated somewhere. Ideally, we would have had a Widgets 1.0 > Core/Umbrella specification where things like this would be defined, > but the P&C spec has taken that role. I think we want our specs to be modular i.e. to separate concerns as much as possible and we should continue being careful to not build unnecessary/harmful dependencies. Based on what you say above, we appear to have met those goals for the P&C spec. Indeed we have talked about creating some type of "uber" spec. I'm not convinced we need it, if all that it essentially says is something like "A widget user agent must implement [P&C], [APIs&Events], [DigSig], [...]". To me, such a statement feels a bit like what I would call a procurement statement (also know as a RFQ) e.g. "I'll only purchase an implementation of widget agent X if it implements specs A, B, C" and that feels like something for the market to decide rather than being codified in a spec. -Regards, Art Barstow > I feel strongly that what specs constitute a widget user agent needs > to be defined somewhere (as well as which are required and which are > not). I'm unsure as to how to proceed here.
Received on Monday, 2 February 2009 12:00:54 UTC