Switch METH_VARGS to METH_O

[movermeyer]

What are you trying to accomplish?

While working on ciso8601, I discovered that things run a lot faster if you define your methods as METH_O (i.e., method with a single object argument) you can skip the overhead of PyArg_ParseTuple.

What approach did you choose and why?

Switched the instances of METH_VARGS to be METH_O instead.

The impact of these changes

Item	Performance of parse_iso8601
master	1000000, 0.2697502020164393 => #270 nsec
This PR	1000000, 0.22211680599139072 => #222 nsec
v2.1.2	2000000, 0.3753614659945015 => #188 nsec

This PR shaves ~48 ns (17%) off the runtime of parse_iso8601.

Testing

Here's the script I used.

import timeit

UNITS = {"nsec": 1e-9, "usec": 1e-6, "msec": 1e-3, "sec": 1.0}
SCALES = sorted([(scale, unit) for unit, scale in UNITS.items()], reverse=True)

def format_duration(duration):
    # Based on cPython's `timeit` CLI formatting
    scale, unit = next(((scale, unit) for scale, unit in SCALES if duration >= scale), SCALES[-1])
    precision = 3
    return "%.*g %s" % (precision, duration / scale, unit)


timer = timeit.Timer(stmt="parse_iso8601('2014-01-09T21:48:00')", setup="from pendulum.parsing import parse_iso8601")
count, time_taken = timer.autorange()
timing = float(time_taken) / int(count)
print(f"{count}, {time_taken} => #{format_duration(timing)}")

What should reviewers focus on?

I didn't bother supporting Python 2.7 in these changes. It's fairly easy to do so, but I noticed that pendulum dropped support for Python < 3.7, so I didn't bother.

I'm not sure why v2.1.2 performed so well in this test. I didn't look into it. 🤷

Pull Request Check List

• Added tests for changed code.
  • Interface didn't change.
• Updated documentation for changed code.
  • Interface didn't change. No publicly facing changes. Just performance.

[Secrus]

Thanks for contributing, but our C extensions are being rewritten into Rust. Superseded by #721