RE: [SVGMobile12] Language for ECMAScript binding is unclear wrt normative criteria

On Sat, 22 Jul 2006, Doug Schepers wrote:
> Ian Hickson wrote:
> > 
> > The ECMAScript Language Binding for SVG chapter says it is normative.
> > 
> > However, I couldn't find any normative text in this chapter. It isn't 
> > clear to me what an implementator is supposed to do with this chapter. 
> > For example, take this definition:
> > 
> >    This method returns a(n) Element .  The namespaceURI 
> >    parameter is of type String. The qualifiedName parameter is of type 
> >    String.
> >
> > What does it mean to say that the parameter is of type String?
> > 
> > Please clarify what the meaning of the statements in this chapter is. 
> > Are they intended to be normative conformance criteria, or 
> > descriptions of some sort of fact which can be derived through other 
> > normative text in the specification? What does it mean to say that the 
> > Node object has a property called parentNode -- is it replaceable? 
> > readonly? enumerable? What happens if it is set to a string value?
> 
> It seems that you might be asking us to replace the word "is" with the 
> words "must be" (and similar terminological changes).  Is this correct, 
> or are you making a more subtle point?

If I tell you to change to using the word "must" then, based on my past 
experience with such requests, you'll do so indiscriminately and end up 
making the spec just as meaningless as now, except with a bunch of empty 
conformance criteria instead of a bunch of meaningless statements of fact, 
which is an even worse position to be in.

So no, I'm not asking you to replace "is" with "must be".

I'm asking you to update that appendix so that it clearly states what a UA 
should implement, in terms of the specification's conformance section, and 
so forth. A vendor should be able to look at this chapter and perform a 
series of tests corresponding to requirements in the specification to 
determine if their implementation is complete and correct. There should 
not be requirements that are meaningless (such as saying that the type of 
an argument is String), and there should be enough requirements that there 
are unanswered questions (such as how to handle cases where a new object 
is created and has its prototype chain set up to include one of the 
prototypes defined in SVG). There should be a clear relationship between 
the members defined in the language-specific chapter and the members 
defined in the language-independent DOM chapter(s), in terms of 
conformance criteria (e.g. stating that a particular method in the binding 
chapter must implement the semantics defined for a particular member in 
the DOM chapter). I'm asking for all the various questions I asked in my 
initial comment to be clearly answered by the prose of the specification, 
along with answers for all the questions of a similar vein that two 
vendors would need to have answered to independently implement the 
specification interoperably.

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

Received on Saturday, 22 July 2006 20:05:36 UTC