Mark Shannon

The number of times we run a benchmark and the duration of run should be independent. The reason for running a benchmark multiple times is to get stable results. How...

> I personally prefer wall clock time. Outputting instruction counts doesn't prevent you using wall clock timings.

One particular thing I would like to know is whether you considered using `__match_args__` to constrict the legal keywords in a class match, so that: ```python class C: __match_args__ =...

I understand that you can't cover every decision, but an overview which documents some of the key decisions would be very valuable, IMO. That's why I asked explicitly about the...

See https://github.com/python/cpython/pull/22917#issuecomment-719596029 I see no justification for sloppy half matching, leaving unmatched variables assigned. Either match a pattern and assign all the variables, or don't match a pattern and assign...

You don't need to erase variables. The values can remain on the stack until the guard passes. Your example could be compiled as: ``` DUP_TOP LOAD_CONST 0 COMPARE_OP < POP_JUMP_IF_FALSE...

> What is it you would like me to do? Nothing for now, but thanks for the offer.

Apple allow runtime code generation for their own code, but not for apps. They presumably allow their browsers to do runtime code generation as well. Whether they would allow Python...

They are trivial to write, but we might as well add them. We should be able to achieve slightly better precision than `analog_write()` can provide.

FTR the DAL provides three variants of `setServoValue()` as well as `setServoPulseUs()`, I intend to provide just `set_servo_pulse_us()`. We can document how to do the trivial maths required.