Re: ATK over a network?

[Milan Zamazal <pdm brailcom org>]

>>>>> "LY" == Luke Yelavich <themuso themuso com> writes:

    LY> I was wondering whether anybody has thought of implementing a
    LY> network protocol for ATK? There is very likely more to it, but
    LY> the idea is escentially this.  A user runs a GNOME/GTK/any
    LY> application that uses ATK on a remote connection.  All keyboard
    LY> and mouse actions are sent normally, but as well as the
    LY> graphics, also send back accessibility information to the
    LY> client, which programs like GOK or Gnopernicus can interpret as
    LY> if the program was running on the local system.

I think the necessary infrastructure is already available.  The
redirection can be done around the border interfaces between
applications and sound output system, i.e. AT-SPI or Speech Dispatcher.

With Speech Dispatcher the simplest way is to redirect the Speech
Dispatcher port on the remote machine to your local Speech Dispatcher
instance, e.g. to invoke the following command on your remote machine:
 
  ssh -L 6560:your.local.machine:6560 -N your.local.machine

This redirects speech output of all remote applications which use Speech
Dispatcher to your local Speech Dispatcher.  If you want to redirect
just a single application, you can set your.local.machine in its Speech
Dispatcher configuration.

I guess it should be possible (at least in theory) to perform similar
tricks with AT-SPI.  Additionally, AT-SPI readers (like Gnopernicus)
should be able to simply talk to a remote AT-SPI instance.  I don't know
any details and whether it's possible with the current implementation,
but since the GNOME AT-SPI server is AFAIK network transparent, there's
no need to invent anything new at least.

Regards,

Milan Zamazal

-- 
I think any law that restricts independent use of brainpower is suspect.
                                               -- Kent Pitman in comp.lang.lisp