

7·
6 days ago@TheMightyCat Oh, this really sounds as a cool update.
It’s a pity screen reader accessibility is not yet a thing on mobile linux at least I can’t find someone making great progress with that.


@TheMightyCat Oh, this really sounds as a cool update.
It’s a pity screen reader accessibility is not yet a thing on mobile linux at least I can’t find someone making great progress with that.
@TheMightyCat Oh yes, thanks for your interest and a nice comment.
A somewhat simplified explanation of linux accessibility stack from an advanced user (not developer) looks like this:
Standard linux screen reader running alongside graphical desktop environments such as gnome or KDE is called orca. It uses a dependency called at-spi2-core for delivering and interpreting accessibility events from different toolkits such as GTK4, QT6 and similar.
Thus at-spi2-core has clients on both the sides. Orca usually consumes events and GUI toolkits and other standalone apps generate them.
For the full picture there are apps that communicate to at-spi2-core directly to provide its accessibility for example web browsers (Firefox, Chromium and webkit-gtk) and libreoffice are doing this.
The fact screen reader is interpreting accessibility events delivered to it through at-spi2-core is a nice thing and it’s a huge part of the puzzle however screen reader has more functionality than dealing with the stream of accessibility events.
For example screen reader needs to be able to review the content by words, by lines and provide different navigation actions.
For that it needs to react to user key presses.
On wayland reacting to keypresses without having a control focused is a security risk thus a protocol extension has been developed which allows screen reader to ask at-spi2-core to coordinate grabbing the keys with the compositor. I don’t fully understand this part so it may be a very rough explanation. However it’s key point to understand that for touch screen operated device it’s not enough to read what has happened in response to some touch screen gestures. We need an ability to review the screen content before executing actions. On different platforms such as android, IOS or Windows it’s implemented in a way so touch interaction model works a bit differently when screen reader is running. This is not yet implemented for linux no matter which form factor.
Technically speaking if I would have an USB keyboard connected to the phone I would certainly be able to use parts of its graphical interface but that would not be practical enough.
Please excuse me for commenting on your announcement thread even adding such a long attempt at explaining my situation. I do really appreciate all the work and advancements you are releasing to the public.
It’s just that I am partially saddened by the fact I can’t experience this linux mobile movement on my own.