[SOLVED] Qtpyvcp - error on startup

Tommy - which preferences menu?

Joco wrote:

I just did a test install using my script on Debian 11 (Bullseye) nonfree, xfce UI per script readme instructions.

Observations:

[1] Did a Developer install.

[2] Install PB. Ran up the sim and on a 1080x1200 screen the UI is displaying perfectly.

[3] Attempted to run "editvcp" on PB UI and found a new dependency issue. Script has been fixed for this. Editvcp works as expected on the UI after this fix.

awesome!  Thanks Joco!!  \o/

are you using a wired internet connection during install? i have had mixed results using wireless so i now always use a wired connection for initial installation and that seems to resolve those issues.

Yep, wired then switch to wifi later.  I've been bitten by that before.

any luck this time?

Promising update.

Loaded MX21, RT6.0, LCNC 2.10 via the instructions in the qtpyvcp thread.

Discovered the screen resolution problem was caused by 'nomodeset' in grub.  Removed that and now all windows are displaying properly, including PB.

That caused latency to jump during testing, so I've been chasing grub and bios settings all day trying to get rid of latency excursions.  I think it's now down to an acceptable level, though not as good as before.

I still get the error about the base_widget in terminal (sim and my config), but I'll circle back to that later.

I also realized after trying to get mine out of estop that PB has been changed a bit since last summer (i.e. the postgui thing from last night).  Some pins in my config no longer exist (qtpyvcp.probe.something) in the latest version.

I think my best bet - sigh - is to start with the PB sim config and start merging my old config in to it and testing as I go.

rodw wrote:

Hmm, cherry picking Joco's install script to skip all the linux-rt and linuxcnc install
github.com/joco-nz/lcnc-bullseye-install...tall_for_lcnc.sh#L99
probe-basic runs fine on my Chromebook which uses Bullseye and a RIP linuxcnc install of V 2.9 once I worked out to home all from the menu. Surely there is a home all button?
 

rod, yes under the dro's, it changes button text to HOMED when homed 

Well, it's fixed.  Mill is moving for the first time in a couple months.

I'm not sure where to start an explanation, or if it would even help someone in the future.

Once I started trying to actually merge the PB sim hal files with my existing config, I discovered some suspicious discrepancies. The PB sim had a postgui file called 'time.hal', and I had both my old postgui.hal as well as a probe_basic_postgui.hal from a different installation.

I'll skip the detective work, but it looks like I had installed the old python 2 version of PB ('master') instead of the required python3 branch.

Fundamentally, I think what was happening was that the 'git checkout python3' line wasn't working - either in the script or in various manual installation instructions.  I probably installed some flavor of linux, lcnc, qtpyvcp, and PB a dozen times and it seems like I never got the 'python3' branch installed properly.  I'm quite sure it was something I was doing wrong, but I just don't have a clue exactly self-inflicted embuggerment I'd managed.

Once I manually switched to the python3 branch and updated everything, it all started working.  Following the same procedure I was able to get my original Odroid-H2 back in business - and the 5.10 rt kernel doesn't have the NIC driver issues - with the exact same config as the HP USFF.  Nice to have a backup PC if I need it.

Now that I can jog everything...back wrapping up the ATC.  Then the 4th axis...

Thank you to everyone for your patience and assistance.

-Ralph

glad you got it sorted out!!