Re: Two finger scrolling

[nia <nia%NetBSD.org@localhost>]

On Tue, Mar 17, 2020 at 05:05:19PM +1030, Brett Lymn wrote:
> On Sun, Mar 15, 2020 at 11:00:09PM +0000, nia wrote:
> > I've just committed a handful of changes to synaptics.c that give me
> > smooth two-finger scrolling out of the box.
> > 
> > I'm curious how many others are using multitouch touchpads and whether
> > you notice any improvement.
> > 
> 
> I will have a look when I get a chance.  Just one question, what happens
> if you place one finger on the touchpad and then place another on the
> touchpad in a different area and move that second finger?  Does the
> cursor move or do you get scrolling?  The answer "scrolling" in this
> situation is bad.  Thinkpads have external buttons wired, some laptops,
> including my fujitsu do not have any external buttons.  That is the
> whole reason for the button emulation code because clicking a click pad
> results in a left button event - that is all you get, to emulate more
> buttons in this situation the region the finger in determines the button
> pressed.  So, to navigate a menu the first finger down determines the
> button then a second finger is used to navigate. 

"Newer" thinkpads (x250, 2015...) have single-button clickpads with extra
wired buttons. Until a few revisions ago these buttons couldn't be configured
to act as regular mouse buttons and defaulted to acting as buttons 4 and 5.

I haven't managed to get right clicks with the clickpad to work either,
honestly. I was not sure how this was supposed to work until you explained
it. To clarify, clicking the pad down with a single finger in the right
area of the pad rarely results in a right click menu either. I worked
around this by adding a knob to recognize the extra buttons as regular
mouse buttons by default.

I've got another laptop with a single button clickpad, and no buttons,
but that's using the ims(4) driver which is significantly less useful.

>
> > The existing approximation (sysctl hw.synaptics.finger_scroll...) wasn't
> > working for me at all, tweaking hw.synaptics.finger_scroll-min either
> > gave me no scrolling or registered normal mouse movements as scrolling
> > 
> > I suspect my fingers are significantly smaller than whoever wrote this
> > code. Was anyone using this successfully, and how does scrolling compare
> > after my changes?
> 
> What can I say? It worked for me.  One thing I have learnt is there is a
> massive variation between laptop trackpads when it comes to reporting
> the W value so values that work on one laptop are bad on others, I am
> surprised that you were unable to get a fscroll_min that would work for
> you.

The number of sysctl parameters with fairly magical values makes configuring
this driver to a usable state a quite confusing affair. But yeah, the
values are not fine-grained enough for me for that particular knob.