Paul Ganssle

> Why does your code parse command line arguments even if it's not the **main** module? @vstinner I don't understand, my code *is* the `__main__` module, it's `pyperf` that is...

> You should avoid using click and pyperf in the same file. Presumably also you should avoid using `argparse` and `pyperf` in the same file, because you'll have the same...

Interesting that the `LocalTZInfo` and `USTimezone` implementations seem to have opposite defaults for ambiguous datetimes.

@mahmoud Yes, the code examples were updated. See this part in the `LocalTimezone` example: ```python # Detect fold fold = (args == _time.localtime(stamp - dst_diff)) return datetime(*args, microsecond=dt.microsecond, tzinfo=self, fold=fold)...

I guess _actually_ running this under Python 3 will be a problem since it pulls in anki as a dependency, and Anki still does not support Python 3. Probably not...

Yeah, I don't see much in the way of Python 3 exclusive stuff out there, so there's a good number of idioms and libraries (e.g. `six`) that help you bridge...

Hmmm.. looking more closely at this, it turns out it's actually an issue with how Anki is packaged, I think: ``` File ".../venv/local/lib/python2.7/site-packages/AnkiServer/apps/sync_app.py", line 34, in from anki.sync import Syncer,...

@dsnopek Yes, agreed that it _should_ be shipping with Anki. As I mentioned in my PR, I think an alternate approach would be to mention this in the documentation about...

`pyproject.toml` is orthogonal to setup.py. This project should definitely add a `pyproject.toml`, but I'm not sure I'd recommend switching it over to a different build backend.

I think the main issue here is that `pandas` has some pretty heavy tight coupling with `pytz` and `dateutil`. They use all kinds of private methods in the `dateutil` time...