Re: Proposal: DBus activation of Accessibility



Hi Matthias,

Matthias Clasen wrote:

> On Mon, May 16, 2011 at 5:29 PM, Frederik Gladhorn <gladhorn kde org>
> wrote:
>> Hi,
>> sorry for cross-posting. I would just like to have this looked at by
>> everyone so we can simply implement it.
>>
>> During the ATK/AT-SPI hackfest it came up a few times that we have no
>> solution to enable accessibility (eg Screen Readers) on the fly. The
>> conclusion was that using DBus would fit everyones needs and make it easy
>> for third partys to adapt.
>>
>>
>> Proposal: DBus activation of Accessibility
> 
> I'm not sure how things work on the KDE side, but for GTK+ based
> applications, this does not really solve the problem. Since GTK+ a11y
> implementation lives in a module that needs to be loaded by all
> applications before a11y tools talk to it. We can load the module at
> runtime, but at that time widgets may already have instantiated no-op
> a11y implementations.

I discussed this at the atk hackfest. This will of course require more work 
for both Qt and GTK+. We know it is not going to work immediately. One idea 
we had for gtk is to create the accessible widgets lazily - as needed. That 
means the dummy no-op a11y interfaces would never be instantiated.
There is a lot to optimize anyway. As far as I understand (without looking 
at any code, so I may be completely wrong) currently tables are 
instantiating lots of objects by accident. That kills their performance. I 
also talked with Mike Gorse about improving the way the cache is handled and 
it seems feasible to implement this.

Short term this means that applications that have been started before may 
not become accessible immediately. But hopefully we will find a solution and 
make this work.

Cheers,
Frederik





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]