Re: [Kde-accessibility] Re: KDE and AT-SPI

[Bill Haneman <Bill Haneman Sun COM>]

On Thu, 2004-09-16 at 13:51, Olaf Jan Schmidt wrote:

> > When the Linux/GNOME accessibility infrastructure was being developed,
> > and when the Linux Accessibility Working Group meeting recommended
> > embracing the GNOME architecture as our standard, DBUS did not exist.
> >
> 
> I don't know which "Linux Accessibility Working Group" you mean. The KDE 
> project is not a member of it, unless you mean the FSG Accessibility 
> Workgroup.

I am referring to "FDAWG", the "Free Desktop Accessibility Working
Group" which was active from 1999 or so through 2003 or so.  This was
the primary place where discussions for free-desktop-wide solutions was
discussed.

At two separate FDAWG meetings (one year apart) in Los Angeles, it was
concluded that at the time all efforts should be put behind the GAP
project, and we wished to rename it to the "GNU Accessibility Project",
though this renaming wasn't done, mostly because nobody had time to
pursue the name change.

> > I agree that in the future alternative implementations of our IDL and
> > protocols should be actively explored.
> 
> I am cc'ing the FAG Accessibility Workgroup chair: Can changes to the 
> protocols be discussed, or was there a recommendation to use AT-SPI as it 
> is and only to discuss "alternative implementations"?

The FAG discussions agreed early on to standardize on AT-SPI.  Again,
this was not something I was pushing, it was a consensus (before KDE was
represented in the Working Group however).  I did raise the issues about
adoption of CORBA and dependencies, and this is why I have been
suggesting that the IDL be the primary normative standard, and that the
standard should allow for alternative non-binary-compatible
implementations of that IDL.

Changing AT-SPI's backend is not something that will happen very soon. 
That is not because we're stubborn partisans, it's just a practical
matter.  I am not objecting to alternative transport protocols, this
idea was always built into the AT-SPI design; however such backends
would need to be based on the normative IDL API.

I think their time will come, and I certainly hope to be a part of their
shared development.  But I don't think the time is now.

Given the difficulty in bringing large-scale assistive technologies to
fruition, I question whether developing KScreenReader and KKeyboard is a
good use of developer resources.  There will be lots of work to do to
get KDE applications and the KDE desktop working with accessibility APIs
like ATK.   Since most platforms which ship KDE also ship the GNOME
libraries, as a medium-term solution I don't see why you would not want
to use the assistive technologies that we already have.

> As far as I know, the KDE representatives in the FSG Accessibility 
> Workgroup have not agreed to any recommendations for or against using 
> CORBA or DBUS for AT-SPI. AT-SPI is on the roadmap for FSG Accessibility, 
> but I understood that no formal decision has been made, and that changes 
> to AT-SPI can still be discussed. If the GNOME Accessibility Project has 
> a different opinion, then please state so clearly.

Of course changes can be discussed, but the FSG/LSB is not in the
business of creating new APIs or interfaces; it recommends and adopts
existing ones.  So any _significant_ changes to the interfaces would set
the standardization process back by years. 

With standards, you basically have the choice of adopting an existing
implementation and design, and perhaps cleaning it up a little bit, or
starting over from scratch.  If you start from scratch, your 'standard'
is at risk of being just as obsolete as the existing ones you rejected,
by the time you are done.

> At the Unix Accessibility Forum, we informally talked about possible 
> changes to AT-SPI (including D-BUS) with GNOME developers. I had the 
> impression that the GNOME Accessibility Project is very interested in 
> cooperating with us and discussing some changes that might be needed to 
> move AT-SPI towards a standard embraced by both KDE and GNOME.
> 
> Bill, when we met at FOSDEM, I had the strong feeling that making 
> accessibility on Unix working is a dream shared by both of us.
> I was also very impressed by the friendly way you helped us finding 
> solutions for ALS. When writing emails, it is far easier to misunderstand 
> each other, especially when one misinterprets the purpose of questions 
> and stated facts. When reading the mail I am replying to, I had the 
> impression that you wanted to say: "The decision for CORBA is made. We 
> can discuss your implementation, but no changes that would mean changes 
> on our side." I am frustrated and disappointed to hear this position, but 
> maybe I totally misunderstood you.

Well, I was feeling very frustrated and disappointed by comments I was
reading and hearing on this forum as well -though not from you Olaf.  I
do hope that all of us reading these lists share the same goal, and this
is why I am so passionate to avoid a "separate but equal" KDE solution
to things which would introduce incompatibilities and new problems for
users.  There are really very few of us available to work on the
issues.  Usually in Free Software multiple projects is a good think, but
I think we are just scattered too thinly on the ground for this to make
sense now.

I have two goals, which I hope are not incompatible.  One is to get a
working Linux/Unix solution that actually solves the needs of disabled
users in the home and in the workplace.  The other is to do this in a
way that shares effort and can be used by all free desktops.  But I
think that addressing the second goal by starting over with a new IPC
protocol would dash the hopes for the first goal's success.  I don't
think it's too early to discuss ideas for AT-SPI-2, but I think that
promoting the existing interfaces and implementation is necessary to the
success of Linux accessibility overall, and in this way necessary to KDE
as well. 

This would deliver a working solution to KDE users in the shortest time,
and then we can work together to remove the more troublesome
dependencies and explore whether there is a better back-end
implementation for free software systems to share.

warm regards,

- Bill

> Olaf
> 
> - -- 
> Olaf Jan Schmidt, KDE Accessibility Project
> KDEAP co-maintainer, maintainer of http://accessibility.kde.org
> 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.0.6 (GNU/Linux)
> Comment: For info see http://www.gnupg.org
> 
> iEYEARECAAYFAkFJjGUACgkQoLYC8AehV8fu2gCdHIDvNi8TLHzwe6YLknEAMm3p
> 6VUAoLssBIk4x/enwR7JLZUBlDNXE3mk
> =AAlD
> -----END PGP SIGNATURE-----
> 
> _______________________________________________
> kde-accessibility mailing list
> kde-accessibility kde org
> https://mail.kde.org/mailman/listinfo/kde-accessibility