Re: Request for Comments: FPWD of Widgets 1.0: URI Scheme spec

Hi Moz,

On Jun 19, 2009, at 10:02 , mozer wrote:
> 1) In the same spirit as WARP, it would be interesting to make HTML5
> reference, an informative one

After careful consideration I believe that you are right — there is no  
direct normative dependency. It is now informative.

> 2) Probably the link between authority and opaque-autorithy should  
> be clearer

I am not sure what you mean exactly. If there is an authority  
component, it is considered to be opaque (i.e. devoid of semantics). I  
went through every occurrence of "authority" in the draft but am not  
sure what I would change to satisfy this comment.

> 3) Update reference to Working Draft 28 May 2009 for Widgets-PC

Done (to CR).

> 4) s/RFC5234/RFC 5234/

Done.

> I'm not sure to fully understand this requirement
>
> [[
> Must be independent of any file system
> Addressing based on this scheme must only map onto Zip relative paths
> and remain independent of any file system on which the widget may be
> stored.
> ]]
>
> Does it mean that, it is case insensitive for example ?

No, it does not mean that it uses the lowest common denominator of all  
file systems (that wouldn't leave us with much :) but that it isn't  
constrained by the FS it's running on and its conventions (e.g. you  
won't get foo\bar on Windows and foo/bar everywhere else). If it so  
happens that an implementation of P+C unpacks (as an internal detail)  
the content of the widgets to a specific on-disk directory, the widget  
URIs must still work the same way (and if that directory is on a case- 
insensitive FS, the implementation is in trouble — but that's not our  
problem).

-- 
Robin Berjon - http://berjon.com/

Received on Wednesday, 2 September 2009 20:58:41 UTC