Touchpanel numerical control bug

[Stijn_Schacht]

When you input a number with a alfanumerical keyboard in a numerical control on a WINCE (not XP) touchpanel which is running only ONE loop: it will be updated.

If you try to input that same number with an alfanumerical keyboard in a numerical control on a touchpanel which is running parallel loops, it will be ignored. However, using the up/down arrows to change or input a numerical control on a touchpanel running parallel loops, will update the control.

Very nasty bug when using touchpanels in operating systems: the operator has no id what hit hem, when he started his machine using default values, because the ones he entered with the keyboard, weren't updated...

there are two attachements:

TEST.vi

TESTHMI.vi
TEST.vi is the main vi to run. It contains one loop, and a reference to TESTHMI.vi, which includes another loop. So essentially, two loops in parallel. You can monitor the parallelness by looking at iter1 and iter2. Normally TESTHMI.vi should be the on top window, so you can experiment with the irresponsiveness of the numerical control, as illustrated above. You should connect an USB keyboard to the touchpanel to reproduce my bug, or you can also use the built-in software keyboard of windows CE.

I have no id wheter windows CE is responsible for this, or wheter it is the CPU that cannot handle keystroke events with multithreated applications?

To test this bug, you will need LV PDA + a touchpanel and a keyboard. I have tested this bug with LV8.5 and LV PDA 8.5 + TPC-2012 and TPC-2006T, I believe it is a general bug!!

If this gets not fixed, I will have to stop using NI touchpanels all together, since they are to dangerous to use.

[smercurio_fc]

Duplicate posting from here, here, here, and here.

[tst]

You do need one thread where this can be discussed. Why not this thread?

The others are either in the BP (apparently by mistake) or not exactly the same.

[Bart De Rouck]

Talked to the client on the phone, about his architecture and adviced him to use a one loop architecture instead of a multiloop parallel architecture.

[StevenA]

Has this problem been confirmed (reproduced) by NI?  Is there a CAR for it?  Most of my applications use multi-loop architecture.  I did find some strange behavior when using the "soft keyboard" to update controls, but never spent the time to track them down.  I found the soft keyboard too cumbersome for my users, so I built my own code for giving the users an iterface to update a control value.  I'm very iterested in the status of this however.

[Stijn_Schacht]

Steven,

don't you have any problems with your own written software keyboard? Do you grab the value from the button, and put it in a control, using the boolean button properties?

If so, I'll try to create my own software keyboard and see if that works. Now that I think of it, it should work since you are forcing the value in the controls from the 'blockpanel-side', and not as a user-input from the front panel. In any case, my applications always use hardware USB keyboards... 😞

[StevenA]

That's correct, using code to update the controls works fine.