Re: [sXBL] Various issues in the sXBL draft -- section 1

On Mon, 28 Mar 2005, Boris Zbarsky wrote:
> 
> The definition of the term "in error" says that error handling is 
> defined by "SVG" (and references SVG 1.1).  It's not clear to me how 
> SVG's error-handling rules should be applied to sXBL.  In particular, 
> would partial bindings end up being attached?  Would an error in a 
> binding make all bindings coming after it in the file not be attached, 
> while those before it are attached?

SVG 1.1 says that as soon as you hit something "in error", you abort all 
processing and display an error message with the page rendered up to that 
point.


> It's not quite obvious, at first glance, whether the namespace URI for 
> sXBL includes the trailing '.' character.  I recomment placing the URI 
> in appropriate delimiters recommended for URIs in plaintext (eg '<' and 
> '>') to make this clearer.

(Ugh, I hate it when people delimit URIs that way. Makes copy-paste a real 
pain in the neck.)

Your point is well-taken, we should probably change the paragraph so that 
the URI is the last thing in the paragraph, and comes after a colon, with 
no trailing punctuation, like this: http://www.example.com/


> The definition of "unless it has already been loaded" presumes a certain 
> amount of control over the HTTP implementation used by the sXBL 
> implemenation (for example, it presumes that sXBL implementation is able 
> to get the post-redirect URI for HTTP redirects).

Yes. This is a requirement for implementing sXBL. (We also assume, e.g., 
that the sXBL implementation has a large amount of control over the DOM 
and CSS implementations.)

The requirement you mention is already a requirement for XML Base, anyway, 
which is a prereq of XBL, so this doesn't add anything really.


> Furthermore, this definition means that any time a reference is made to 
> a resource via an HTTP URI and the server prevents caching (fairly 
> common for all sorts of script-generated and CMS-generated things) the 
> sXBL implementation MUST make an HTTP request to determine whether the 
> URI might now result in a redirect.  This seems rather undesirable.  
> I'm not quite sure what this section is trying to accomplish, but the 
> proposed solution has some unpleasant side effects; perhaps there is a 
> better solution?

Could you give a more specific example of what you mean? I agree the 
section is poorly phrased, and would like to define it better, I just 
don't see how to do so.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Tuesday, 29 March 2005 00:27:50 UTC