Re: GDM accessibility sans AT-SPI

[Brian Cameron <Brian Cameron Sun COM>]

Henrik:

> Another controversial post to g-a ...
>
> I'm currently looking at GDM accessibility and it strikes me that there 
> is a strong case for doing this without using AT-SPI. The themed version 
> currently does not work properly with the AT-SPI features and on the 
> plain greeter version there is still a fair amount of configuration 
> required.

For some things, avoiding AT-SPI is a good idea, I think.  Especially
for magnification where it might be easier to just make it easy to
increase/decrease font size in the GDM windows/dialogs.  Note that
this sort of feature should work for gdmgreeter, gdmlogin, gdmsetup,
gdmchooser and all dialogs popped up by these programs.

I suppose it might be possible to code an on-screen keyboard directly
into GDM, but this might be more work than you think.  Note GOK supports
"dwell" mode so that it works for users who can only manipulate a single
button.  Making an on-screen keyboard that supports the same sorts of
disabilities that GOK supports might be tricky.

Text to speech would probably be hard to get working with gdmlogin,
gdmgreeter, gdmsetup, gdmchooser and all pop-up dialogs.  While it might
be possible to do something that would work okay without AT-SPI, the
danger is that users might end up in a situation where they don't know
what is going on with the GUI.  The advantage of AT-SPI is that it
works better for following the focus and context of what the user is
doing.

> Both the AT-SPI framework and the assistive apps are complex things that 
> will need some work to get working Just Right at the login. It also 
> takes some time to load. AT-SPI is great for global desktop access since 
> adding access to every single app would be silly. However, GDM is *not a 
> desktop app* and also has a simple and predictable interface so it makes 
> sense to look at other options.
>
> I've written a spec describing a login system with built-in access support.
> https://wiki.ubuntu.com/Accessibility/Specs/GdmAccessLite
>
> This may not be the right way to go but I think we should consider it 
> before starting work on fixing the current model.

I think it might be possible to get things working for some disabilities
without at-spi.  However, we probably still want to support the option
of using at-spi in GDM for users who really need some AT program.  Note
that in the future additional AT programs might become available that
users migth want to use in GDM.  So you can't assume that the current
set of AT tools are the only tools that people will want to use in GDM.
For example, perhaps dictation solutions that use AT-SPI will become
available in the future.

Brian