Re: issues with X Window

[Paolo Molaro <lupus lettere unipd it>]

On Mon, Aug 23, 1999 at 09:55:27PM -0600, L. C. Robinson wrote:
> > The first requirement means that our tool should work with an
> > application without changes to the application itself.
> 
> This sounds like a rather unlikely ideal (which fact you comment on
> below).  It seems to me that such a tool is really trying to add
> text capability to an app, which a synth can then convert to sound,
> with the added restriction that you cannot directly modify the app
> itself.   Such an approach will no doubt force the programmer to go
> through some rather tortured contortions to get things to work,
> with less than optimum results.  This makes sense with proprietary
> software, where there is no alternative (and vendors are making
> such apps available for linux),  but I wonder if it makes as much

No one forces the programmer to do anything that is not good design
(and using gtk+:-) and no one restricts you to make modifications
to the app. The programmer doesn't have to do any contortions to
get things to work. A gtk application provides us with all the info
needed to do a good audio interface without the programmer ever
thinking about it! Some custom widgets or interfaces need a bit more
work as I said.
Let's take gnumeric as an example: when I tried it the first time with 
GSpeech *I* was surprised to see it working quite well. I could
open a file, edit the cell's formulas, saving in a different format
etc. The things that I couldn't do (change the current sheet and read
the *result* of a formula) are being addressed by Miguel. The first
problem is annoying even for a sighted user (not being able to change
sheet with a keyboard shortcut) and I'm sure would have been addressed 
sometime anyway. The second change was needed only because gnumeric
uses a custom, optimized display engine for the cells of the sheet.
Note: you will not be able to master The Gimp all of a sudden
thanks to GSpeech, but you will be able to use a spreadsheet and some
other apps that is now impossible to use.

> sense with OSS; might it not work better to just contribute patches
> to the code base of such apps to make text mode versions possible?

This is not always possible, requires a lot of work for some apps
that are inherently GUI oriented and the majority of users require
graphical interfaces.

> There are already many apps that have text and GUI counterparts
> for linux and *nix: this tradition can be encouraged, probably with
> much better results than trying to make everything work under the
> GUI, which isn't necessary under linux.

We are not trying to make everything work under the GUI, we are trying
to make accessible applications that will never be ported to a text 
interface. Don't take me wrong, it's very welcome if someone ports the 
apps to work in text-mode, but don't hold your breath.
The primary interface for a blind will always be the text-mode one:
you got emacspeak or some advanced screen reader for the emacs haters.

> Correct me if I'm wrong, but, if usablity enhancements have been
> added, won't they be application specific additions to the screen
> reader tool (which will be necessary in most cases)?  If so, won't
> you have to upgrade your tool anyway whenever you upgrade or add
> applications?  What happens in the M$ world relative to this?

Look, a menu will work even when you upgrade the application.
The same happens for all the standard interface components (and
gtk+ provides a lot of these).
The way GSpeech is designed prevents a *lot* of binary incompatibilities.

> Has it proved to be feasible to make truly application independent
> screen reader tools for any GUI environments anyone knows of?

I think it's feasible and I'm looking for your help and criticism
to make it real.

> Do such tools have application specific modules or "drivers" with
> them?

GSpeech already provides the support to load application specific
modules. The point is that most applications don't need them.

> Now, I note that that Paolo has made it clear in another message
> that text mode apps are better for most uses, and he seems to be
> aware of the problems (and understands the programming issues far
> better than I): I was just concerned that some posters in this and
> related threads seem to have some illusions that GUI speech tools
> can be made to solve all access problems, with one clean, sweeping
> approach.  

The last thing I want is to spread illusions. GSpeech is not the 
solution for all the problems, I think I said it clearly before.

> Correct me if I'm wrong, but I suspect that even when
> such tools sort of work, they will generally be kludgy work
> arounds.  In the OSS culture, we can do much better (and already do
> so, for most uses).

Ok, take it this way: most programmers are not aware of the issues 
involved designing the interface in a way that makes it easy for the 
blind to use. With GSpeech you can try the apps and report the
usability problems you find (like in the OSS culture), without it
you are stuck with an unknown world.

Alessandro replied wonderfully to your other mail.

Looking forward your useful comments,
	lupus

-- 
lupus debian org
Se devo emigrare in America, come mio nonno, prendo il tram!        [F.G.]
If I have to emigrate to America, like my grandfather, I'll take the tram.