[Usability] Re: The importance of good mouse acceleration

[Michael Toomim <toomim ocf berkeley edu>]

Jonathan Vaughan wrote:

> Please let me know what your thoughts are on this issue, and the best way to
> proceed.
Have you talked to the X maintainers yourself?

As GSR pointed out, there is already code in X that lets you use a 
decent polynomial acceleration algorithm.  The problem is that A) this 
isn't documented, and B) you can't set the constant multiplier factor at 
runtime in a reliable way... you can set the "Resolution" parameter 
(which sets an equivalent linear multiplier) in /etc/XF86Config, and 
(with certain mouse drivers) at runtime using an X extension (XInput I 
think), but the interface is not well supported... it didn't work with 
the drivers for the USB mice that I used, anyway.
What's needed is a good interface at the desktop level that uses this 
polynomial acceleration algorithm instead of the default crappy X 
acceleration.  But that needs better support from X, and will only work 
with X servers that support it.  I'm not familiar with the recent 
political changes in the XFree86 organization, but I bet there would be 
someone in one of the forks willing to accept patches to improve this 
situation, if you found the right person.