Re: gtkhtml2 vs. gtkhtml1

[Bill Haneman <bill haneman sun com>]

On Wed, 2002-09-18 at 13:04, Michael Meeks wrote:
> 
> On Tue, 2002-09-17 at 16:03, Bill Haneman wrote:
> > > Bill, why do we need DOM?
> 
> 	Currently there is a AtkDocument interface that exports a single
> method:
> 
> 	gpointer getDOM (); // or sim.
> 
> 	Which seems to me not the best interface to build the requirements
> around ;-)

That API is irrelevant to the current discussion, mostly.  It's there in
case we figure out a reasonable way of exposing DOM info or a pointer to
something that we can do DOM operations on.  Unfortunately only
normative C++ and Java DOM bindings are available, and though the DOM
IDL is standard, I have been told by very authoritative sources that
exposing the DOM via CORBA "doesn't/won't work".

 
> > CSS is very helpful for accessibility, but there are other ways of
> > reformatting.
> 
> 	I couldn't agree more; there should be no need for access to the DOM if
> we can provide good accessible interfaces for the document. Indeed,
> having the DOM tree in parallel to the Atk tree seems to cause far more
> trouble than it's worth, and a far more limited set of information to
> boot. Oh, and then we'd need a CORBA interface for the DOM for it to be
> useful to AT vendors.

Sorry, take it up with the w3c (I've tried) ;-/

> 
> > However the w3c accessibility User Agent Guidelines require that DOM be
> > made available by HTML and similar content viewing agents.
> 
> 	The question is - is that a sensible requirement; and does it fit in
> with the way Gnome a11y works ? or is that a minimum requirement for
> things that have no other form of accessible interface.

It's a "requirement", period.  There is at current no means of meeting
the guidelines without satisfying that requirement.
 
> > Perhaps a future solution would be to use a gecko or Moz=based viewer
> > for such content in "accessibility" cases.
> 
> 	Ultimately it's a huge, long term (and profoundly unnecessary to my way
> of thinking) task, so - it would seem that the minor task of providing a
> DOM interface to FooHtmlRenderer is small in comparison ;-)

I don't agree really; what w3c requires is that a DOM interface be
available.  It can in fact be an in-process interface, their
requirements say nothing about exporting it out-of-process, etc.

But it does appear that ultimately that is what will be required, for
this and other projects.  I am not arguing that it's a a particularly
practical requirement or even the best solution for users, I'm just
relaying the requirement, and it is one of the few hard-and-fast,
explicit accessibility international "standards body" requirements we
have.

(please, don't shoot the messenger ;-)

-Bill
 
> 	Regards,
> 
> 		Michael.
> 
> -- 
>  mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot
>