Re: [g-a-devel]Porting At-SPI to Mac OS X

[Bill Haneman <Bill Haneman Sun COM>]

I'd like to clarify a couple of points:

On Thu, 2003-10-09 at 19:38,
gnome-accessibility-devel bernard-hugueney org wrote:
Peter Korn wrote:

...In practice, this first step would
> > probably run into several hurdles. These include the Macintosh X
> > window manager (which may not implement all of the modern calls
> > that at-spi uses - I just don't know one way or the other),
> 
> I did not realized that at-spi talked to the window manager :-(

at-spi doesn't talk to the window manager; it only talks to the X server
and the bridges, as you assumed.  However at-spi does require, at this
time, the XKB and XTest extensions.  I am not sure whether they are
available on OSX.

> I thought it was only talking to the X server (for events) and with 
> bridges gail and gnome-java-bridge. I hoped I would only (!) need to 
> have a gnome-cocoa-bridge and a gnome-carbon-bridge.
> 
> > potential ORBit interoperability issues when it comes to use with
> :-( I did not realized that either

I don't foresee ORBit problems; in fact I think the ORBit stuff has
already been ported via the Fink project.

> > Java and OpenOffice.org, some minor serial port wierdness for
> > Braille (and of course porting BrlTTY will be another issue),
> > potential USB wierdness for GOK input devices, and a host of
> 
> I must confessed that I am not interested by input devices.

Input device support is key to successful use of GOK.  The only Java
issue is that for OpenOffice.org accessibility support, a Java 1.4 JVM
is required.
 
...
> > The second big step is integration with Macintosh applications
> > (using either Carbon or Cocoa).  There is a Carbon accessibility
> > API (see
> > http://developer.apple.com/documentation/Carbon/Accessibility-date.
> >html) and also one for Cocoa (see
> > http://developer.apple.com/documentation/Cocoa/Conceptual/Accessibi
> >lity/).
...
> > There are two big issues to solve here.  #1 is reaching API parity.
> >  When I last looked (quite some months ago), there was still a lot
> > missing from the Apple accessibility APIs that is in at-spi and
> 
> I think that I would be satisfied even if complete parity was not 
> achieved.
> 
> > that our two assistive technologies take advantage of.  #2 is
> > figuring out a bridging mechanism. Do you use CORBA?  Do you use
> > some other IPC mechanism for marshaling objects?
> 
> I don't use anything yet :-) I thought I'd use CORBA.

I'd recommend leveraging the Fink/ORBit work (unless I am mistaken about
ORBit2 being already ported to MacOS X).

Good luck with your investigations Bernard; in principle I think this is
quite achievable provided the Cocoa and Carbon bridges can be
successfully created.  

best regards,

Bill

> > Are you serisouly interested in taking on this project?  It would
> > be a tremendous thing to have.
> 
> Well, I don't want to disappoint you. I'm quite satisfied with Linux 
> for at-spi. Nevertheless people have expressed the desire to use my 
> application (GUI usage tracing thanks to at-spi) on Mac. I hoped it 
> would be easy enough to be tackled as a project by students in my 
> university :-) (2-3 students, 4 months)
> It's a kind of lottery : you can pick the best students or the worst 
> :-(
> The experience might still be interesting.
> I must also say that the people wanting to use my application was 
> ready to give some (Mac knowledgeable) help in the developpement.
> 
> Do you have any estimate of the manpower needed for a basic (no input 
> devices, either cocoa or carbon (wichever is easier) bridge without 
> complete parity) ?
> 
> I guessed I could at least target gnome apps on Mac Os X. The bridges 
> would be done at a later stage.
> 
> Bernard
> _______________________________________________
> Gnome-accessibility-devel mailing list
> Gnome-accessibility-devel gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-accessibility-devel