Q: Initialization procedure

[leloup314]

Is there a particular reason for the following syntax to be necessary?

from basil.dut import Dut

# Instantiate my_device
my_device = Dut(my_config)
my_device.init()  # Additionally call an init method on the instance manually

Why not call my_device.init inside the instances __init__?

[themperek]

I do not remember anymore but there was a reason for this.
We sometimes manipulate things before actually connecting to hardware.

[leloup314]

From my subjective perspective it looks like 90% of the use cases do not need to do this.
Maybe it would be nice to have something like a default argument Dut(auto_init=True) that automatically calls the init as last action inside the __init__ if set. If not wanted, set auto_init=False

[themperek]

Dut(auto_init=False)

This should be OK.

[leloup314]

Well this will not change anything for anyone ;) I would add Dut(auto_init=True) and issue a DeprecationWarning when people do

my_device = Dut(my_confif, auto_init=True)
my_device.init() 
# Prints DeprecationWarning("Manual call to 'init'-method is deprecated if 'auto_init=True'. Use 'auto_init=False' if manual 'init'-call is required" )

This way change could actually happen. I can implement this easily if you want to test

[leloup314]

But of course this would break some stuff. My point is to design the API for the edge-cases is not the right approach. Can you tell me where you do something in between the initailization of the Dut and the init call?

[themperek]

In general OK but not sure if this is a good idea to do now. Should accumulate more braking changes?

[themperek]

Maybe not best example: https://gitlab.cern.ch/silab/bdaq53/-/blob/development/bdaq53/system/bdaq53.py

[leloup314]

The only line where something happens in between initialization and manual call to init is for the clock generator. Everything else looks fine to me.

[leloup314]

But yes, maybe we can wait and have a more thorough look at this, maybe accumulate other braking changes.

[cbespin]

Maybe not best example: https://gitlab.cern.ch/silab/bdaq53/-/blob/development/bdaq53/system/bdaq53.py

Just to add, since tj-monopix2-daq is almost exactly like bdaq, this will also be affected. No problem for me to adapt, though.