[3.5.1] Macro restarting

[tristanryerparke]

I've been doing a bunch of experimentation on an SBC driven system with, sd jobs, macros, pause, resume, M400, and dsf-python.

I've encountered some issues and general unreliable results when triggering commands and macros from various code channels, especially when interacting with pause/resume during a file. So I preceded to try and create some minimal examples that showcase my issues. When trying to create a control example (without all the channel stuff) I came across this issue with macros.

Steps to replicate:

• Fresh 64 bit duetpi install on RPI4, sudo apt-get update,sudo apt-get upgrade (everything is at 3.5.1)
• Upload of my config.g
• Change LogLevel to debug in /opt/dsf/conf/config.json
• Reboot
• Home
• Upload of a gcode file test_custom_code_in_file.nc.g (added .g) for upload
• Upload of a macro move_x_macro.g
• Running of test_custom_code_in_file.nc produces expected effect, machine moves down in the Y axis, enters macro, moves over on X axis, finishes macro, and then continues to move down in Y. sudo journalctl --unit duetcontrolserver -f gives the following output during: normal_file_w_macro_nopause.log.g
• While running test_custom_code_in_file.nc, the user hits pause in dwc while the x axis is moving. The macro completes and the x axis moves out to a value of 400.
• User presses resume, and the macro is re-executed the X axis goes back to around 0, and repeats the motion out to 400. normal_file_w_macro_pause.log.g

As far as I know, unless the macro has an M98 R1 (which mine does not have) , the file should resume directly after the M98 call. I tried adding an M98 R0 command to the beginning, but got the same result.

For macros that toggle physical systems, i.e tool-change-like actions, a repeat like this can cause a crash.

My eventual goal here is to figure out how to pause a file without the fear of a macro being repeated.

Then I can move on to to diagnosing my other issues.
Thanks
T

Edit:
I tried going back to 3.4.6, and things functioned in the way I would expect. The macro did not restart after resuming.

[droftarts]

@tristanryerparke This sounds somewhat similar to the issue reported here: https://forum.duet3d.com/topic/35544/resume-g-activating-twice-with-fw3-5
It's on the list for @dc42 to have a look at. I'll alert him to this thread.

Ian

[tristanryerparke]

@droftarts Thanks.
Is it possible to go back to 3.5.0-beta4 to temporarily avoid the issue while still having some of the 3.5 features?
I couldn't find it on the rrf/dsf apt package index.

Edit:
I built and installed rrf/dsf/dwc 3.5.0-beta4 but the issue persisted, so I had to stay at 3.4.6 to avoid the macros restarting.

[dc42]

@tristanryerparke if you haven't already, please upgrade to 3.5.2 and report whether it fixes this issue.

[tristanryerparke]

@dc42 On 3.5.2 I'm noticing some different behavior.
This time when the user presses pause while the move_x_macro.g is running, the x axis stops at X=300 and the machine enters a paused state. The macro does not seem to restart upon resuming.
The final line of the macro is G0 X400 which indicates the last movement command is being skipped? I tried adding a G0 X0 to the end of the macro and repeating with the pause, which made the machine stop at X=400. Repeating the G0 X0 did not help either.

EDIT:
I tried adding M400 or G4 P1 at the end of the file as workarounds for the last command issue, but this made the restart issue re-emerge.
Can provide logs again if need be.

Thanks,
T