Attention is currently required from: fixeria, neels.

laforge has posted comments on this change by neels. ( https://gerrit.osmocom.org/c/pysim/+/39741?usp=email )

Change subject: [1/6] personalization: refactor: drop ClassVarMeta use ......................................................................

Patch Set 7:

(1 comment)

File pySim/esim/saip/personalization.py:

https://gerrit.osmocom.org/c/pysim/+/39741/comment/f3b91bd8_214049bb?usp=ema... : PS7, Line 155: class SdKeyScp80_01(SdKey): I'm still somewhat doubtful that this is an improvement. You're turning something that was a very nice single-line syntax into something that now needs three lines to declare something. The original syntax was almost like a table, with columns listing each key_id and key_usage_qual at the same positinin of each line...

What I really fail to see is some kind of argument in the commit log how the new approach is better than the old one.

I really don't want to spend weeks or months arguing about this. I just want to understand what is the improvement here. "Not using a metaclass" doesn't seem like a qualitative improvement of the code. What was the problem caused by the metaclass? Why is it desirable to not use one? The questions are not rhethorical. I tend to use metaclasses quite a bit in pySim and elsewhere, and so far nobody has argued in code review that anything is wrong about this.