Is this a bug or a "feature"?

[ITmaze]

Not sure if this is a bug, or if it's a PEBKAC error.

FreeCAD v1.0.2 [Rev: 39319] on MacOS. I was unable to locate the Snapmaker post processor, so on 9 November 2025, I added it from this repo to my ~/Library/Application Support/FreeCAD/Macro directory where it seemed to be working fine, until it didn't.

I created a toolpath that crashes my A350 controller. I have looked at the controller logs and I see a weird Java error after attempting to load the gcode file called 0-mounting_caphead.cnc - I'm including this here as it's the only record I have of what's happening on the Controller. The gcode file is 4.9K with 144 lines.

141511,1970.01.01 00:02:21.511,INFO,SC,Click file directory false path /20251203/0-mounting_caphead.cnc
141580,1970.01.01 00:02:21.580,DEBUG,SC,Route: Create PreviewFragment
141774,1970.01.01 00:02:21.774,DEBUG,SC,preview fragment view created...
141784,1970.01.01 00:02:21.784,DEBUG,SC,Update result: work speed = 100.0, estimated time = 0.000000s
141799,1970.01.01 00:02:21.799,DEBUG,SC,Reset overrides.
141908,1970.01.01 00:02:21.908,ERROR,SC,io.reactivex.exceptions.UndeliverableException: The exception could not be delivered to the consumer because it has already canceled/disposed the flow or the exception has nowhere to go to begin with. Further reading: https://github.com/ReactiveX/RxJava/wiki/What's-different-in-2.0#error-handling | java.lang.ArrayIndexOutOfBoundsException: length=1; index=1
141910,1970.01.01 00:02:21.910,ERROR,SC,	at io.reactivex.plugins.RxJavaPlugins.onError(RxJavaPlugins.java:367)
141911,1970.01.01 00:02:21.911,ERROR,SC,	at io.reactivex.internal.schedulers.ScheduledRunnable.run(ScheduledRunnable.java:69)
141912,1970.01.01 00:02:21.912,ERROR,SC,	at io.reactivex.internal.schedulers.ScheduledRunnable.call(ScheduledRunnable.java:57)
141913,1970.01.01 00:02:21.913,ERROR,SC,	at java.util.concurrent.FutureTask.run(FutureTask.java:237)
141915,1970.01.01 00:02:21.915,ERROR,SC,	at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:272)
141916,1970.01.01 00:02:21.916,ERROR,SC,	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1133)
141919,1970.01.01 00:02:21.919,ERROR,SC,	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:607)
141920,1970.01.01 00:02:21.920,ERROR,SC,	at java.lang.Thread.run(Thread.java:761)
142537,1970.01.01 00:02:22.537,INFO,SC,Application started!

After the controller restarts it comes back with a connection error that can sometimes be resolved, and the files cannot be seen in the folder where they are stored. I also see Module errors in the logs. I updated the firmware from v1.14.1 to v1.20.3, but that made no difference. Rebooting makes the connection error go away and the files reappear, but the problem recurs if I attempt to load the gcode file, or any from the same split output (it generates five files).

The only "odd" item I can see is that there is an M3 S10000 in the output. Even if I manually replace that with M3 P100, or M3 S255, (based on the Snapmaker Gcode Wiki) the controller still crashes.

I cannot see any other "changes", other than a feed rate that's been changed from F450 to F600 in the G3 commands:

G3 X-101.585 Y-100.000 Z-0.692 I-1.585 J0.000 F600.000

I then attempted to use the current post processor, by removing the existing one, adding the current one, removing the __pycache__, restarting FreeCAD and now I get no gcode output at all. The error I see is this:

12:22:15  Running the Python command 'CAM_Post' failed:
Traceback (most recent call last):
  File "/Applications/FreeCAD.app/Contents/Resources/Mod/CAM/Path/Post/Command.py", line 188, in Activated
    post_data = postprocessor.export()
                ^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/onno/Library/Application Support/FreeCAD/Macro/snapmaker_post.py", line 891, in export
    return self.snapmaker_process_postables(filename)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/onno/Library/Application Support/FreeCAD/Macro/snapmaker_post.py", line 614, in snapmaker_process_postables
    gcode = self.export_common(objects, filename.format(name=name))
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/onno/Library/Application Support/FreeCAD/Macro/snapmaker_post.py", line 835, in export_common
    if not Path.Base.Util.activeForOp(obj):
           ^^^^^^^^^^^^^^^^^^^^^^^^^^

module 'Path.Base.Util' has no attribute 'activeForOp'

Other things to note:

• The output file name is set to %D/%S.cnc, which works fine for the 2025-11-09 post processor version, but not for the 2025-12-05 version. I attempted to hardcode a name, which made no difference.
• Between the two versions the toolhead name changed, I updated the arguments in FreeCAD accordingly.
• I also recreated the tool paths from scratch, but that didn't seem to make any difference.
• The USB drive with the gcode is formatted with FAT_32 and it still loads older files that were generated by the 2025-11-09 post processor on the same FreeCAD model, but with the tool paths generated a week or so ago.

My Questions:

• Does anyone know why the Snapmaker post processor doesn't appear to be installed on MacOS -- I'm using homebrew to install it.
• What is causing the 2025-11-09 version of the post processor to crash the controller.
• Why does the 2025-12-05 version fail to create any output -- other than the error.

As I said, this might be a PEBKAC error, or there might be something else going on. Happy to test and figure out what needs doing. I didn't lodge a bug report, since I'm not sure what the bug is, or if it's a bug or user error at all.

No doubt I've forgotten to add some salient information, let me know. I have purposely not (yet) added the gcode files, since I'm not sure if that's the actual issue or not, since they all render fine on various online gcode visualisation tools, including using the Luban Workspace.

Onno

[clsergent]

Hi!

Here are my very first comments.

The postprocessor doesn't ship with FreeCAD:

You're the second one to report this (the first one was using Ubuntu). I don't know why this is happening, I have checked FreeCAD sources and it's ok there. I'm not using the homebrew version but weekly releases, I'll check that and report to the FreeCAD team if necessary.

The toolhead name has changed

This breaking change was implemented by a FreeCAD contributor to allow adding non-CNC toolheads (laser, drag knife, etc.). While I’m usually reluctant to break backward compatibility, it’s an interesting argument, and since it has been merged into FreeCAD, let’s stay synchronized.

Differences between 2025-11-09 and 2025-12-05 versions

Could you provide the commit number for each version? I’ve added several new functions recently; maybe it’s related to one of them.

%D/%S.cnc bug

I have not been able to reproduce this issue on macOS. It works fine on my computer so far... Weird

GCODE M3

You can add the --spindle--percent argument to use % rather than RPM. If there is still a M3 Sxxx please report because it would be a bug.

GCODE G3

While G2/G3 always worked on my machine, I know that the snapmaker firmware doesn't handle these commands well (listed as unverified in the wiki). Maybe your java bug is related to this although it seems incoherent. There is a new --split-arcs argument in the FreeCAD API but it's not in the releases yet. Could you try remove all G2/G3commands and see if it still crashes your machine (please be cautious as I don't know it's consequences on you toolpath, except that it won't be ok).

I don't see any other problem so far, and the postprocessor is working fine on my machine, but I can't cover every use case. If none of the above suggests work, could you share your GCODE and CAD file to help me figure out what is going on?

[ITmaze]

This is going to take a moment. I just discovered that my Snapmaker A350 can now suddenly read those files. The only thing I changed was to copy a new file into the same directory on the thumb-drive.

Using a serial connection I discovered that all my backlash and carefully measured axis steps per units have vanished.

I also noticed that the Snapmaker didn't want to boot up the first time, I suspect because the EEPROM versions no longer matched.

I'm writing this down, just to make sure that I'm not going nuts.

I'll reset everything, re-calibrate, then start again and see what I learn...

Not sure if this qualifies as a PEBKAC or if this is a Snapmaker "feature".

I'll update this thread when I have more.

[clsergent]

I would go for a Snapmaker feature...

[ITmaze]

I hear you .. I'm inclined to agree, that said, there's still weird interactions between the Post Processor and FreeCAD, so that's why I'm not giving up.

I'm kicking myself that I relied on the EEPROM store for those values, never even occurred to me that a firmware update would cause this to no longer work, behind the scenes Snapmaker updated the Marlin version as well. Interestingly the machine was rebooted several times before I started this thread, but it seems that being switched off overnight (and the potential that the file copy "helped"), seems to have made an incremental improvement.

The only good thing to come from this is that I'm now writing down the procedure to do the whole calibration for the next poor sod in this position -- likely me -- last time it only took half a day, so perhaps this time it will be quicker -- ha!

o