> Following the addition of 'mouse button autorepeating', it'd be nice if the
> driver detected affected mice and set up them so that they worked as
> expected "out of the box" (i.e., as advertized by the manufacturer).
> 
> The attached patch does this (for my specific mouse model) by defining
> a quirks table in ums(4) and setting the repeat rate if it finds the mouse.
> I don't think this is possible to do at the wsmouse(4) level because it
> cannot identify mice, can it?  I mean, each bus has its own way to
> describe devices (e.g., vendor/product in usb)... unless we used a string
> with the mouse's full name, in which case wsmouse(4) could do it.  Mmm.

what you want to implement is "make autorepeating default for
buttons which are normally used for some special purposes,
for which autorepeating is often expected."  right?
is it appropriate to call it a quirk?

i'm not sure if it's desirable to put this kind of functionalities
into kernel.

> Note that the same automatic setup can be applied to "emulate 3rd
> button", when implemented.

i don't think that whether 3 button emulation is desirable
is only a property of a device.

YAMAMOTO Takashi