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

Re: Iggdrasil, a new amazing screenreader



>>>>> "john" == john doe <johndoe65534@mail.com> writes:


There are a couple of reasons you might want a screen reader with a
different architecture than Orca.

Orca's architecture makes it relatively easy for Orca to block.
It's written in python, and if anything in Orca does something CPU bound
while holding the global interpreter lock, the entire screen reader
blocks.

Similarly, if Orca makes blocking interactions with something like dbus,
or blocks waiting for some other application to respond to it, Orca can
block.

I don't know about the mix of first or second class of problems.
I do know that when interacting with large tables or large DOM trees,
Orca can block, disabling accessibility for all Orca-read applications.
I've definitely found the combination of Orca/Firefox to be problematic
particularly when interacting with large DOMs like build logs, or
sometimes app.element.io.

There are various things that Orca could do to be better about this.
And in fact, Orca has gotten better over the last couple of years about
only hanging Firefox rather than the entire system.  (Although I did
have a situation just a few minutes ago where ORca hung entirely until I
waited for ever or quit Firefox)

I haven't looked at what Orca actually does internally, nor the efforts
to reduce blocking over the years.  If someone did look at that and
concluded that a new architecture--one with better threading or first
class co-routine like structures (Think Rust or go) was appropriate, I
wouldn't disagree with them.

I hope the community gets behind a small number of solutions.
But if there's an actual good reason for trying something new,
particularly when the efforts to fix what we have are stalled, that
might be the right answer.

--Sam


Reply to: