I've seen some other people posting similar issues, did these problems start with the switch to GTK3 or might it be closer connected to the switch to Preempt RT or a later Debian version?

I'm not really competent to determine exactly what's going on, but I did look through old GH issues - and there are a number of similar reports over the years.  They go back to 2.7.

However, there may be two (or more) underlying problems which present the same - more or less - undesired behavior.

I get the sense that one of the scenarios involves switching tabs or otherwise changing window focus during a continuous jog.  At the very least, something other than just pressing/releasing a keyboard key is happening.

The other scenario is what the OP reports - no change in window focus, no pop-ups, nothing but a continuous jog even though the key is released.

The two scenarios may not have anything to do with each other beyond the end result.

My suggestion for the OP would be to turn off keyboard auto-repeat in Debian settings and see if the problem continues.  If the problem goes away this may point to a Debian change rather than an LCNC change.

This is becoming useful for debugging a running LinuxCNC instance. I still have some features to add and bugs to fix before releasing an installer.
Up-to-date source code is available at the link on my first post.
image missing https://git.disroot.org/nwe/halflow/src/branch/main/media/Screenshot_2026-06-08_07-45-10.png

video:
git.disroot.org/nwe/halflow/src/branch/m...a/halflow-tour-1.mp4

Ja, erstmal die bestehende Konfiguration auf den neuen Rechner nehmen (ohne Ethercat)
Bei Problemen die dmesg posten (immer sehr aufschlussreich)
Wenn es keine Probleme gibt, dann ethercat mit rein.
Bei Problemen wieder die dmegs posten. 

I haven't tried it, I'll try it tonight, so I'll let you know, thanks a lot

    If you have kept up in the news of what has been going on with 3D printing then you know it concerns more 
then just putting plastic on a plate with a computer. It concerns anything CNC additive or subtractive. 
Well now at least in the the USA there seems to be a name to at least one of the ones responsible. To be clear I
am not making this accusation , but it sure feels reminiscent of the 1920s alcohol prohibition. 

Ok, du meinst ich soll den Ethercat Treiber nicjt laden ? 

Ich suche alles raus wenn ich heute abend zuhause bin. 

Da ich die Maschine aktuell noch ohne servo erstenmal zum laufen bekommen mochte habe ich diesen noch garnicht eingebunden. 
Noch ist nur die mesa Karte und die sxhrittmororen angeschlossen. 

 

I'm running the standard Axis UI, with a Glade VCP side panel and a tab. (But I'm sure I was not using or touching the mouse at the time that the key sticking occurred, because I press the jog keys with the same hand that I use the mouse with.)

PS.  This really has the feel of a bug to me.  The older versions of LinuxCNC I had been using (2.7) running on RTAI kernel versions of Debian and Ubuntu, had never exhibited this kind of dangerous behavior.  I've seen some other people posting similar issues, did these problems start with the switch to GTK3 or might it be closer connected to the switch to Preempt RT or a later Debian version?

I will look at the config but that is actually a separate application
Did u try axis 2 on a small screen it should fit I think it removes a slider or 2 for convenience

Doing my first steps and trying to configure and understans how this works and seems found one tiny BUG:
Debian Bookworm
NIC card configuret for 2 VLANs native for data, .20 for MESA with Layer3 switch.

# The loopback network interface
auto lo
iface lo inet loopback

# Management interface
auto eno1
iface eno1 inet static
    address 192.168.4.50
    netmask 255.255.255.0
    gateway 192.168.4.1
    dns-nameservers 8.8.8.8 1.1.1.1

# MESA P2P interface (.120-.121)
auto eno1.20
iface eno1.20 inet static
    address 192.168.1.122
    netmask 255.255.255.252
    post-up sysctl -w net.ipv6.conf.eno1/20.disable_ipv6=1

Seems like its trying to reach a wrong folder during startup..
Note: Using POSIX realtime
sysctl: cannot stat /proc/sys/net/ipv6/conf/eno1/20/disable_ipv6: No such file or directory

the right place is: /proc/sys/net/ipv6/conf/eno1.20/disable_ipv6

Be careful with the set/clear model, your idea is M6 sets extended limits to be able to reach the tool change? That means in normal conditions your limits are not including the tool change area, this could be a problem if you want to jog there to verify anything, could even prevent you from loading gcode without errors if preview sees the tool change moves, also think about emergency stopping during an M6, you will not get to run the clear command. Just pointing those things out, I would consider having open limits and only restrict them when running and not doing a tool change, that should get you what you are looking for, without any drawbacks.

Build errors? Can you share your current code? Not many people will have the time to read 10 pages of an 8 years old thread...
Be more specific about what you are struggling with.

Your core assumption is unfortunately mistaken, it cannot be done by only modifying the INI file, coresim.hal is only 3 axis, the fourth axis will f-error as it is not receiving any feedback, this is expected behavior.
My suggested quick fix, in your INI file after

HALFILE = cooling.hal

add a connection to a local file like

HALFILE = fouth-axis.hal

then create the file next to the ini file with contents

net Apos joint.3.motor-pos-cmd => joint.3.motor-pos-fb

This will allow you to move your fourth axis, don't expect the rotation to look like what you expect I think A axis is rotation around X, so it might look unexpected depending on what your fourth axis does, but it should work.

If you are looking for a deeper sense of accomplishment, read the coresim.hal file and research and understand what it is doing, and why it is needed, that will get you much more satisfaction and future know how.

Hi,

du solltest den neuen Rechner erst mal ohne Ethercat testen.
Wenn alles Stabil läuft, nimm die Ethercat Komponenten mit rein.
Poste dann die Ausgabe von dmesg.
Zusätzlich: Welche Version vom Ethercat hast du installiert?
Zusätzlich: Liste mal deine Ethercat Hardware auf   (Karte, Kabel CAT??)
Zusätzlich: Wie hast du den Servo angeschlossen und wie sind deine Kabel verlegt?



 

Basically, the control works with the software-inverted signal, but this sometimes results in more or less lost steps.
Since it works perfectly with the Mesa card, the solution using RS422 or the ACT drivers would probably be more reliable.
What components would I need for a reliable solution, ideally with multiple channels per axis?

I've been using CAMotics for some time to visualize and check G-Code programs. I'm now hand coding some simple programs with a fourth axis (XYZA). I've found CAMotics doesn't recognize the rotary axis and also doesn't recognize the global variables used in LinuxCNC, like #<_variable>. So, in a weak moment of delusion, I thought I could install a simulation version of LinuxCNC on my office computer to verify G-Code files. Great idea, I've now spent 2 solid days on it but I'm "almost" there.

I have a VirtualBox on my office computer for Windows/Fusion 360. I've been able to install a simulation mode LinuxCNC 2.9.4 (bookworm) on the Virtual Box. I could not find a four axis (XYZA) simulation version so I've installed this "sim.axis" XYZ version. It simulates fine for 3 axis (XYZ). I'm trying to add a fourth axis to this version. I think this can be done by modifying only the axis.ini file? But, with the mods I've done, the simulation doesn't move in the axis display and then faults with "joint 3 following error".

If anyone has insight in this area I'd sure like to find out what I'm doing wrong. Attached is the axis.ini file with mods noted. If you can offer assistance I'd be grateful. Thank you.