iaccessible2 and linux (was Re: Lubuntu and Accessibility)

Eric S. Johansson esj at harvee.org
Sun Jun 12 11:01:20 UTC 2011

On 6/12/2011 2:40 AM, Isaac Porat wrote:
> Hi
> My comments were not a criticism of at-spi but rather the need were possible 
> to unify accessibilities standard across platforms for the simple reason that 
> if software vendors have to worry about one accessibility stack it is better 
> than two or three, as Linux as very small user based it is always left behind 
> by the main vendors.

in my opinion, any criticism of the current accessibility world should not be 
focused on just cross platform but on how they don't really meet the needs of 
the disabled, application developers, or accessibility interface developers.

I start from the principle that accessibility is defined by what the user needs, 
not what the application but accessibility interface vendor is willing to give. 
For example, nuance doesn't give me what I need (a speech user interface was 
sufficient discoverability) therefore, they don't provide sufficient 
accessibility. At the same time, I a assert that individual is responsible for 
their own accessibility because what the user needs depends on what they do and 
the nature of their disability. There's no way that any application or 
accessibility vendor can possibly provide that level of customization at a price 
that individual can afford.

The second starting point is that trying to replicate a GUI or extract 
information from a GUI is a failed proposition. Speech user interfaces (spoken 
or heard) have entirely different structure. A blind person using a web 
application should here a small number of things essential to operating the 
interface and not all of the junk around the application unless they ask for it. 
A spoken interface is a wide and shallow interface where control the scope makes 
the same or similar command do different things.

My personal objection to most of the current work is that they seem to ignore 
speech recognition entirely or so cripple the interface as to be useless for 
anything but direct text dictation in a very limited way.

The current accessibility tool kits are replicating the same mistakes I've seen 
inaccessibility environments since I've been disabled. I think a better 
solution, is for the application to export the data and operations available via 
a GUI and accessibility tool to reveal all of its controls and data so that a 
customizable framework can drive both application and accessibility tool to 
provide the accessibility interface necessary for the user.

the model I propose reduces the cost for an applications developer to provide 
accessibility, makes it possible to split boundaries between which machine has 
the accessibility device, and which machine has the application as well as 
applications within browsers. This model also makes accessibility cheaper, 
easier to validate, customize, and support than the current models.

this is still theoretical as I don't have any hands with which to write code.  I 
wouldn't turn away a volunteer to help me with making a prototype. the basic 
idea is sound enough that others I have bounced it off of think it's worth 
exploring. All I need is someone with hands.

--- eric

More information about the Ubuntu-accessibility mailing list