Re: Proposal: Conformance Statement

Denis Anson wrote:
> 
> Ian,
> 
> This language says that compliance can only come through native
> implementation, which would leave out third party AT as a means of
> compliance.  I don't think that this is necessarily a good idea.

Denis,

I'm not sure which language you're referring to. The proposal
reads:

>          2.Satisfy those checkpoints natively (i.e., no additional
>            software is required) unless the checkpoint explicitly
>            indicates that it may be satisfied through communication
>            with other software.

This means (and perhaps it needs editorial work if not clear enough)
that the WG will have to choose which checkpoints must be implemented
natively and which may be implemented through an interface. Thus,
conformance depends on some hybrid, which the WG must establish.

Furthermore, conformance for ATs is defined, in exactly the same
way: some combination of natively implemented checkpoints and
communication through an interface. 

This conformance proposal implies two (challenging) tasks for the WG:

 1) Establish two subsets of checkpoints
 2) Decide which ones must be implemented natively.

I hope this addresses your questions.
 
> One of the themes that has gone through the document in techniques is to
> fully implement and expose the DOM as a means of conformance.  But that
> implies that a third party device would use the information provided by the
> DOM to provide access to the web.  The requirement of native implementation
> would make exposure of the DOM a non-issue, since the browser must provide
> direct access to all priority 1 items.

As mentioned above, conformance is not defined as native implementation
of all checkpoints. Thus, discussing communication through an interface
- The DOM - remains a vital subject for consideration.
 
> By allowing third party AT, we also give the option of such technology also
> implementing priority 2 or 3 items, even though the mainstream browser does
> not.

The following provision of the conformance proposal can be
tightened up to allow this if the WG desires:

>        Even for those checkpoints that must be satisfied natively, 
>        desktop graphical user agents should make information 
>        available to other software through standard interfaces. 

Currently, it reads "should". Making it a "must" would ensure that
other software could implement the other checkpoints that are
not required for conformance. The WG must decide if it wants
to strengthen this requirement.

 - Ian

Received on Tuesday, 2 February 1999 10:02:36 UTC