[BRLTTY] Two problems with Pegasus display from Deininger

[Dave Mielke]

[quoted lines by Sebastian Humenda on 2010/04/03 at 10:47 +0200]

>The problem remains. 

This is the problem where you sometimes appear to be missing a keyboard release 
event when using a keyboard key table?

I personally work with a keyboard key table (in my case, -kkeypad) active all 
the time, in part so that I'll encounter any problems which may remain within 
that code. In spite of this, I haven't encountered a problem of the type you're 
describing.

I'm assuming that the key release event does eventually arrive as you're not 
reporting indefinite repeating. It must be, therefore, that it just seems to 
arrive late on occasion. I'm suspecting, therefore, that you may actually be 
seeing something that I have observed, which I believe to be a problem within 
the Linux kernel combined with a common problem among typists (myself 
included).

The problem within the Linux kernel is that newly created keyboard devices 
(well, event devices, but I'm trying to keep it simple here) autorepeat at a 
miuch higher default rate than the keyboard itself. Brltty is impacted by this 
because it has to create a new keyboard device in order to echo back to the 
kernel the key events that it doesn't handle itself. What this means from a 
user perspective is that you get a higher repeat rate when brltty is running 
(and you've specified a keyboard key table) than you do when it isn't running.

The typist problem is that people, often unknowingly, are a little slow in 
releasing keys, especially when typing quickly and also especially when the 
next character is typed by the other hand. I'm no expert in this, but, as 
mentioned above, I myself am guilty of doing this. I suspect our brains put 
more attention on getting the next key rpessed quickly than they do on 
releasing the previous key. Those of us old enough to remember having used 
manual typewriters were forced to face this as, whenever we'd type quickly, we 
often ended up with massive hammer jams. With the much softer touch of today's 
electronic keyboards, I suspect this problem is much more significant.

So, if I'm right, what you're observing is explainable. If fast typing is 
causing you to be a little slow at releasing keys, and given that Linux is 
defaulting the keyboard device that brltty is echoing back into to a higher 
repeat rate, it may simply be that the key repeating is kicking in much earlier 
than you're expecting it to.

I've already been trying to figure out a good way to solve the higher default 
repeat rate problem. So far, though, I've been unsuccessful at getting the 
right current value back from the kernel. I'll look into it again.

>Should I provide for this problem also a log file? I
>suppose this one would be huge. 

Yes, it'd be huge. Please try to figure out if my guess above bears any 
resemblance to reality first.

>I don't know whether brltty supports timestamps in the log file, but it would
>be easier to track down the error in the log file, when it has just happened.

It doesn't write timestamps right now, but, yes, it'd be nice if it did. I've 
just been reluctant to add yet another command line option.

>For the other problem with the first routing key I started brltty and
>clicked once on the first and then once on the second routing key. The
>same I repeated in a text editor. I attach the file below.

The log file shows both keys being pressed. Insofar as routing keys are 
concerned, it shows five keys: first, second, first, second, first. The log 
further shows that the core is indeed seeing all of them and interpreting them 
correctly.

-- 
Dave Mielke           | 2213 Fox Crescent | The Bible is the very Word of God.
Phone: 1-613-726-0014 | Ottawa, Ontario   | 2011 May 21 is the Day of Judgement.
EMail: dave at mielke.cc | Canada  K2A 1H7   | 2011 Oct 21 is the End of the World.
http://FamilyRadio.com/                   | http://Mielke.cc/bible/