Re: Exposing object model to applications

At 10:33 AM -0500 4/14/99, schwer@us.ibm.com wrote:

>This should be integrated into the techniques section (6.1) that I
>submitted earlier this week. It appears to follow the general techniques I
>had specified in that the assistive technology can register itself for DOM
>access through a system registry mechanism as a DLL with a specified entry
>point that can be called with a reference to the DOM.


If we are trying to round out the picture, two possible other
techniques which should be added would be:

1 - out-of-process access to the DOM
   yes it can be done, it is not necessarily slower, and in many cases it
may be the only option available

2 - write your own UA
  a.k.a. pwWebSpeak


Can people think of any other solutions, or does that cover the
spectrum?  I believe #3 and #4 would be as already described....

3 - launch the entire AT inside the address space of the (UA) browser
[see Rich's prior email]

4 - launch some part of the AT, a piece of stub code, a DLL, a Browser
Helper Object [special DLL] as Rob suggested, etc., inside the address space
of the (UA) browser [see Rob's prior email]


>What is not clear here is whether the IE Object model, IUnknown, will
>guarantee full implementation of the DOM interface. My assumption is that
>it does. I strongly support Mark Novac's concern that as a minimum an
>assistive technology should be guaranteed that the current approved W3C
>specification for the DOM be supported.

I've asked the same question regarding IE's support of the W3C document
object model, more specifically, their level of support for the HTML Level
1 DOM.

>I had spoke with Chuck Opperman at CSUN and he said that access to IUnknown
>was provide but not a publicly supported function. Is this not the case?
>Might this change? Is access to IUnknown reentrant?

Received on Wednesday, 14 April 1999 16:24:27 UTC