Onyx Boox as monitor

[mdp]

Originally Posted by sogaiu

May be the folks with interest in using the Max 2 with the Raspberry Pi will be interested too.

They certainly will!

Originally Posted by sogaiu

I think I saw the "touching the bottom left corner" functionality working still -- is that correct? If so, is that distinguishable from a click / touch intended for the desktop? I guess there is an answer in the .smali

The corners still fully work, they must - they are part of the proficiency in using the system. The video shows switching from Normal to A2 to Normal in more occasions, plus the full screen refresh (which appeared to me double, I hope I do not have to correct it?), and the NW corner opens the menu.
What happens is, if you tap in the corner you also send a click. You may want to tap where you have no side effect if you intend to trigger mode change etc.; and if you have to click on something in those corner areas, you will also trigger the effect on the "Monitor" mode. I did not have any better solution.

Originally Posted by sogaiu

P.S. The double-clicking looked slightly tricky. How was that for you?

It does ; ) You can see it also in the video.
Certainly, you cannot do it with a finger; you can do it with the pen but you have to be precise.
The two clicks of the double-click action must be in very near coordinates...
I was already surprised that I can nail the windows title bar well with the finger!

[sogaiu]

Haven't worked on the support for 'extended' modes of operation, but I've incorporated other things mentioned.

- the model is obtained via adb
- the resolution is guessed using mouse.screen_size()
- the log lines being considered start at the most recent using -T 1
- scaling(?) factors have been introduced that incorporate device and host resolutions
- the CPU usage has been reduced by sleeping
- code has more structure and commenting
- profiling (via yappi) has been added but disabled to not burden folks with additional requirements

[sogaiu]

Regarding double-clcking...

Originally Posted by mdp

Certainly, you cannot do it with a finger; you can do it with the pen but you have to be precise.
The two clicks of the double-click action must be in very near coordinates...

I found that selecting first and subsequently attempting to double-click seems to work better here. I was able to do it with the finger that way a few times.

[sogaiu]

Originally Posted by mdp

plus the full screen refresh (which appeared to me double, I hope I do not have to correct it?), and the NW corner opens the menu.

Ah, I think I see two refreshes too.

Verified that the menu is brought up via the NW corner -- nice!

What happens is, if you tap in the corner you also send a click. You may want to tap where you have no side effect if you intend to trigger mode change etc.; and if you have to click on something in those corner areas, you will also trigger the effect on the "Monitor" mode.

Is it true the corners are now smaller? At first I didn't have much luck triggering the various behaviors but as I moved closer to the corners, things seemed to work better.

[mdp]

Originally Posted by sogaiu

Is it true the corners are now smaller?

Yes, I set them large (256px) in the first mod to prevent "Hey, it does not work for me",
but returned to contained amounts (100px) in the second since there are now side effects with the touchscreen.

[sogaiu]

Originally Posted by mdp

Yes, I set them large (256px) in the first mod to prevent "Hey, it does not work for me",
but returned to contained amounts (100px) in the second since there are now side effects with the touchscreen.

Thanks for the clarification.

BTW, along the lines of using this with the likes of Raspberry Pis, what do you think about on-screen keyboard functionality? Does it seem feasible?

[sogaiu]

Regarding on-screen keyboards, I guess for *NIX the following sort of thing might work:

http://list.xmodulo.com/florence.html

[mdp]

Originally Posted by sogaiu

BTW, along the lines of using this with the likes of Raspberry Pis, what do you think about on-screen keyboard functionality? Does it seem feasible?

I see - you mean, when you connect to systems that cannot have a virtual keyboard of their own, such as teletype consoles.
There surely may be ways to use Android ones, but it is not trivial.
Really, either you have a physical kb connected to the HW, or VNC is what you want in such cases.

If you are connecting to a Desktop OS, instead, there are probably plenty of virtual keyboards available for install.

But real case scenarios... Suppose you are connecting it (HDMI, of course) to a tablet - ok, then you may /extend/ its desktop and place a virtual keyboard enlarged to be as big as the screen on the original. You view on the Onyx, you type on the source tablet.

[sogaiu]

I was just thinking that not having to connect a keyboard to a Raspberry Pi to use it might be nice -- but if it has some Desktop / X environment running, as you say, a virtual keyboard should work fine.

[mdp]

Originally Posted by mdp

The two clicks of the double-click action must be in very near coordinates...

We actually have a way to facilitate double-tap as double-click:
-- store details the last "down (= 0)" event: x, y and t(imestamp);
-- if down and this_ts-last_down_ts<1000ms and abs( this_x-last_down_x )<50px and abs( this_y-last_down_y )<50px then this_x = last_down_x , this_y = last_down_y
You get the pseudocode.

[sogaiu]

Thanks for the hints -- I'll try to see what those might lead to.

[sogaiu]

Ok, attached should be an attempt to improve the double-clicking experience.

For tweaking double-clicking, see the lines:

dbl_click_delta = timedelta(milliseconds=1000)
dbl_click_res = 50

[Randy11]

Originally Posted by sogaiu

... May be the folks with interest in using the Max 2 with the Raspberry Pi will be interested too ...

[mdp]

Originally Posted by sogaiu

Ok, attached should be an attempt to improve the double-clicking experience

I took a little time to get some stats:
-- the double tap time can be very subjective (affected by age?), but mine kept lax varied from 100ms to 350ms, with an average and median of 250ms. I think that to keep the "timedelta" value at 800ms will be safe.
-- the tap displacement in pixels, I think I well guessed it with 50:
7% of the deltas were beyond 40px, which means a success rate in double-click action of 85%;
4% of the deltas were beyond 50px, which means a success rate in double-click action of 90%; a few were beyond 60px. I attach a chart.

EDIT: a bit more empirically:
30px → 80% ; 35px → 85% ; 40px → 89% ; 50px → 93% ; 60px → 95%

EDIT2: I removed the former image because it was shamefully wrong. The distribution revealed to be a logarithmic distribution (10000 cases will err within 20, 1000 within 40, 100 within 80, 10 within 100 ... ).

[mdp]

With the current progresses, using the Max2 as monitor is coming just short of amazing, productivity wise, realistically speaking.

The problem remains of that flashing.
It happens with colours that once pre-processed by the Max2 become akin to #808080 and very close neighbours: it happens with #7f8180, but not with #7f7f7f or #828282.

Now, the posted example should make it evident...