7i73 LCD shows rubisch

In Andys example, the component "threads" is setting up the real time thread
( it defaults to 1 ms AFAIK)

oups yes,
thank you I learned something else

OK tried Andys test.hal script on LinuxCNC 2.8 and 2.7: (with adjustments for host FPGA type, I used a Ethernet card)

2.8 Works (blinking underline cursor disabled)

2.7 Does not work (cursor never disabled= hash on screen)

I think that there were some cleanups in the LCD component that were not merged to 2.7.
This is born out by using the 2.8 LCD component source on LinuxCNC 2.7 (this fixes the cursor issue)

I suspect this may be a long standing bug that was not noticed due the long response time of LCDs
which would make the scattered cursor hash nearly invisible.

In that case, the answer is to download this file:
raw.githubusercontent.com/LinuxCNC/linux...hal/components/lcd.c
and then compile and intall it with

Thank you very much to both of you. Congratulations, it's always good news to locate a bug and fix it.
I will compile the new component and try as much as possible.

I have one last question, what is the point of setting the timing of the servo-thread (1 ms) in the ini file ?
No matter how valuable it is, when I start a session, and I watch the timing of the servothread with halmeter
(fperiod) it always has the same value 0.15ms with my configuration.
just like the time of each loaded rtcomponent .

The component times are how long they take to execute (in CPU clocks)

Take a look at motion.servo.last-period for the CPU clocks per servo thread invocation

Perfect, both types of OLED board work well !

Hello,
the scatering cursor problem appears again with 2.8.1
if you haven't seen my other message, 2.8.1 limits the size of the xml file linked to PyVCP
and thus refuses to start

gaston48 wrote: Hello,
the scatering cursor problem appears again with 2.8.1

That seems rather odd, as I don't think that the lcd component changed.

github.com/LinuxCNC/linuxcnc/commits/mas...hal/components/lcd.c