I want to use QTAxis, is it meant to be used?

Is there a technical reason why we can't open gcode before the machine is enabled?

No.

In qtvcp/lib/toolbar_action around line 86 - change to:

elif action == 'load':
#STATUS.connect('state-off', lambda w: widget.setEnabled(False))
#STATUS.connect('state-estop', lambda w: widget.setEnabled(False))
STATUS.connect('interp-idle', lambda w: widget.setEnabled(True))
STATUS.connect('interp-run', lambda w: widget.setEnabled(False))
#STATUS.connect('all-homed', lambda w: widget.setChecked(True))
function = (self.actOnLoad)

Thank you for that adjustment, axis itself allows loading files without activating the machine.

Going back up the stack of things I asked... regarding the reload button you asked if I had a pref file path set

First, I do not, but verbose mode says That file does exist and contain the preferences, so I think this should be working fine.

Second, if I set the value without a full path like qtvcp does not like that and fails to start with
Stack trace   in case you'd like to examine this.

Ok I have pushed a fix for the 'PREFERENCE_FILE_PATH = qtaxis.pref' error.
Thanks for the detailed report

I think reload works for graphics, but it does not reset the view or clear the plot lines.
I loaded chips.ngc, then edited the tool number and reloaded - it used the new tool number.
I think it should at ;least clear the plot lines.
I see the gcode display does not reload with the button push though.

I also didn't know linuxcnc's interpreter will always reload the file when you press start.
I don't think that's a very good idea but I guess I should add a check for changes to warn the user.

For now I will have reload clear the plot and fix the gcode not updating.
Thank you again.

Actually I think linuxcnc has a problem where files are accessed on disk for the entire time the gcode is running. I have recently copied new ngc files onto the machine while the interpreter is running and things break.

I do understand that this is my fault, but it should probably get locking/tempfile/something safety check to make sure the interpreter doesn't do anything dangerous as a result of replacing an ngc file on disk while it is running.

Just a quick note on my supposition above, I'm not sure of this... and I'm not trying to lob it out and say it's a bug report, but the whole problem of editor/preview/interpreter loading of the file seems to be disconnected to me and I want to learn why.