[PATCH libinput 0/6] Middle mouse button emulation

[Bill Spitzak]

On 04/15/2015 02:46 PM, Peter Hutterer wrote:
> On Wed, Apr 15, 2015 at 07:32:01AM -0700, Bill Spitzak wrote:
>> Is this being done for physical mice with two buttons? The comments seem to
>> indicate not, which seems kind of pointless: if a client has to emulate it
>> for a physical mouse it will be emulating it for touchpads as well.
>
> not quite, there are two parts to it: configuration and the actual
> emulation. You may get both or just the emulation, if you have both you can
> turn emulation on/off otherwise it's always on. in the current patchset:
>
> * if a device has left+right but not middle, emulation is enabled by
>    default but no configuration option is present (i.e. cannot be turned off)
> * if a touchpad has a middle button, no emulation or config option
> * if a non-touchpad device has a middle button, emulation is disabled by
>    default but can be enabled through the config option

I would check on Windows or Mac I think it is pretty common that 
emulation works whether or not there is a real middle button, maybe with 
the ability to turn it off.

>> Delaying a press to see if it is a middle click is also probably annoying.
>> It should send the down event for the first button, then the down event for
>> middle when the second button is pressed (ie the client will see it as a
>> left+middle or right+middle click).
>
> yeah, no. you now have the onus on the client to guess whether it needs to
> filter the left click or not.

If there is a fixed latency between a button being pressed and the 
result (because it is waiting to see if it is a middle button emulation) 
then it is going to look slow to the user. This is especially annoying 
because single button clicks are probably more common than middle button 
emulation.

Alternatively the latency will be so short that it will be difficult to 
get an actual middle mouse button emulation.