Re: [orca-list] Accessibility of apps

[Jason White <jason jasonjgw net>]

MENGUAL Jean-Philippe <jmengual linuxfromscratch org> wrote:
> It's a common assessement: if free software isn't accessible, it's not a
> screen reader problem, but the app which doesn't send good info to at-spi.
>
> ok. How to fix this? Explaining, bug reporting, etc. 


I think it will require a dedicated group of people who are comfortable with
accessibility APIs and who can look at a project's source code, identify the
cause of a bug and fix it quicker than the project's own developers can. A bug
report can then be submitted with a patch that fixes it. The project's
maintainers can help to improve the patch, but the point is to provide a patch
which has been tested as a starting point, not just a bug report.

I think there's more than enough money being made from Linux to finance what
I've just described, and there might also be willing volunteers who can
assist. I would have expected the regulatory situation to have prompted Linux
distributors to pool their resources in this area and to dedicate effort to
it, but it hasn't happened yet, presumably because we haven't seen any
litigation yet. Is that what it's going to take? I hope not, but I fear it
might be.

The second aspect to this is automatic regression testing and much more
rigorous testing in general before projects are released.

The third component involves documenting the accessibility APIs more
effectively for developers and publicizing them.

I'm sure there are people who have better ideas than I do - consider the above
as opening a conversation.

Lastly, be aware that there are accessibility bugs and regressions on every
operating system. The proprietary software world has more resources, including
more resources to dedicate to accessibility, but nobody, so far, has
consistently mastered the art of bringing the quality of the experience for
screen reader users even close to the quality of the experience enjoyed by
users of the visual interface. I've used every major platform over the last
several years, a number of them for work purposes, and it's the same familiar
story of accessibility bugs to a greater or lesser extent across all of them.
The Linux console is the only environment I am aware of in which a screen
reader user can install an application and be very confident that it will be
accessible - not completely confident, of course, but highly so. That's
entirely due to its being a text-based environment. The same issues apply to
Web applications as well.

A lot of attention has been given to accessibility APIs, but the software
engineering problems of implementing them correctly and consistently haven't
been addressed yet.