On Sat, 26 May 2001, Dave Morse wrote:
> 1) When X is rendering like mad, it mouse and keyboard events are not
> delivered in a timely manner.
If you hook to an input that has real kernel-managed file descriptor
and poll on it, like linux-evdev, this will probably help a lot
in this situation.
> 2) No way to distinguish between real user keystrokes and typeomatic
> repeat.
For keys, we have separate events for press vs repeat vs release.
For mouse buttons we only have press vs. release. Best in either
case to keep track yourself which keys are down, and treat press/repeat
events both as a repeat, if the key is already down.
> Is this something that gii can address? Is there a simple top down
> explanation of how its supposed to work somewhere?
Basically any input you load should be no different from others
except you may have to pass them options to help it initialize
on some systems, what options depend on which input is being used.
See the doc/inputs.txt file in the libgii source. The strings are
passed via the GII_INPUT environment variable when you are not
using LibGGI, and they look something like this:
GII_INPUT=linux-mouse:/dev/ttyS0
(Evan: note, when you decide what to do with inputs.txt nowhere
in there is the "input:val" or "input:-option=val" actually shown
or any examples given, which is bad since noone can figure out
how to actually write an option string from what is there.)
There are HTML docs on the API (as soon as the latest batch of
DNS/webserver problems have been worked out) which should help
you figure out the usage; in the meantime,
ls /usr/local/man/man3/*.3gii to get a list of manpages for LibGII
--
Brian
