Re: GNOME A11y: where do we need to improve? (Want input by 25-Jan)

["Steve Lee" <steve fullmeasure co uk>]

On 26/01/2008, Francesco Fumanti <francesco fumanti gmx net> wrote:
> I don't understand exactly what you mean by general purpose ATs; what
> would be the non general purpose ATs? In fact, in my view the
> fonctionalities of the "plugin" would be available sessionwide

We may be saying the same thing but using 'plugin' differently,
especially from your comments below. I only meant that something like
prediction or TTS has use with other ATs. If it is a plugin that only
works with one AT then we the work will be repeated and plus if a user
uses 2 ATs (screen reader and OSK say) then there will be duplication
and conflict. If such work becomes a general 'service' usable by
various ATs then we are in a much better position. The closest I think
we have now is magnification which is a service and ATs like Orca
drive it. One possible solution would be to have a AT plug in
architecture that all ATs could use. However the UIs would vary
greatly between ATs.

> And now the following occurs to me: should I not say systemwide
> instead of sessionwide!? For example what about GDM? Say a user
> installs an AT; should that same AT not automatically be available at
> the login-screen, but not necessarily activated.
[snip]
> Does a plugin have to announce itself to the session? If that is the
> case, I would suggest that the plugin should not only announce itself
> to the session, but also to gdm.

Fully accessible login is a holy grail to aim for ;-)
You have some good ideas here. I guess the term 'plugin' is one that
has various meanings and in the scenario you describe I wonder if
there is another GNOME term as I thought 'plugins' are generally
application specific (and that is what I thought you meant). 'Applet'
perhaps, but that usually something in the panel?

> For the moment I don't know whether such a modular architectures with
> "plugins" makes also sense for the other ATs available.

It might be possible to design such a modular infrastructure with care
and cooperation. We don't have a large number of ATs/a11y features.
Then again it might be something available already with a OS and it
just needs the right UI designed.

> Do I get it right: the suggestion is to use a numerical computation
> to find the completion and prediction word? I am curious: Is there
> any theoretical model supporting the suggestion?

No, I think it was just it was a bit of 'out of the box thinking' that
if you entered an expression it could get evaluated as the result.
Compared to predictive text that gets 'evaluated' as the word.

> >A script plugin would be useful for power users.
>
> What kind of scripts are you talking about? Does the script provided
> by an application (or plugin) not depend on what that application
> expects?

In the context of single application plugin architecture I just meant
that power users could want a level of customisation greater than
available through GUI config options. That opens up the program for
community as well through sharing scripts. 'Scripts' could be purely
declarative (like HTML, say). I thought a plugin would provide the
environment to run scripts within its GUI (cif the ipython plugin in
accerciser) . The same could apply in a system pluggable
infrastructure, though shell scripting is a possibility and yes  each
program would expose its own API. I was only envisaging one language
or format (e.g python) not bindings to many.

> Will (for example) Windows user even look for free software? I don't
> know, but I assume most will not. I fear that they have the reflex to
> walk into a shop and buy.

That seems so in the AT world or at least for those who purchase and
support users. They appear to be suspicious of FOSS and reluctant to
support more systems. On the other handI get the impression that many
end users of general software expect PC stuff to be free (e.g tucows,
download.com etc) but are not always clear on the differences between
freeware/shareware and FOSS and what the FOSS guarantees are.

> Priority should (in my opinion) be given to make the switch possible
> for users that want to switch, by providing them the AT they need.
> And if it is of outstanding quality, it might even attract new users
> who learn about it on specialised forums and lists.

If other users share their good experiences and so help reduce the
'fear' of change. Even better, other users help them make the
change....

> (Besides AT, there are also other things that can hinder the switch,
> for example years of emails stuck in a client that cannot be moved
> properly to gnome,...)

Ah but that is an example of where having cross platform programs
removes silos and platform lock-in. Using Open standards (e.g. email
DBs) and harmonisation of APIs across platforms means developers have
less to do to avoid this sort for problem for users.

> Of course, if there are enough resources to make the product
> outstanding and cross platform, it will be better. (If making it
> cross platform requires only little more resources, it might also be
> worth doing it; but again: it should not be at the expense of the
> quality of the product.)

Agree.  and as you say having the superior product on Linux might make
people vote with their feet/mouse/switch.

> However, I want to remark that I am not an evangelizer; I don't know
> what makes users really switch; so I might also be wrong.

Nor do I. Though in education it is often the presences of someone who
is an enthusiast and has the skills and time. It would be valuable to
find out. Perhaps a survey?

 Steve