linuxcnc.command().mdi() vs .program_open()

Can anyone tell me the difference(s) between calling an ngc routine with .mdi("o<myfunction> call") and .program_open("myfunction.ngc")?

Follow-on:

I'm working on a python program to check the levelness of my table.

I have several o<word> functions that are currently attached to buttons on the gmoccapy vcp, one to set zero_z, one to make the measurement and one to reset the spindle back to zero_z preparatory to the next measurement. All work as expected.

The python program is an attempt to automate the process measuring the four corners and the center of the table, then record the values into a spreadsheet. The hope is that overtime I can create run charts for each point to evaluate accuracy and precision.

Currently, the application moves to the Z height and sets zero_z calling the first .ngc function as expected. Next it moves the spindle to the first corner and calls the measurement.ngc. The spindle is lowered to the touch plate, but fails to move up and back down for the final measurement. It appears as though the function is failing to execute all of the commands before returning to the application complaining that the spindle failed to reach the touch plate.

I'm wondering if calling an o<word> function is limited to executing just one command when called by c.mdi("o<word>")?

Perhaps o<word> commands need to be processed with c.program_open. However, your previous response indicates that the code would be loaded but, not executed. 1) How do I get the file to execute after opening it, 2) Will loading a program preempt the execution of the running python program?

Any hints to get around this SNAFU?

Regards,