Torsten Wörtwein

and for multiple non-overlapping bar plots on the same figure `bar shift=0` should be set

> Appears to be some typing failures The typing checks are now green :)

Convincing pyright/mypy to infer "inherited" methods sounds unlikely https://github.com/microsoft/pyright/issues/3812. Using mixin sub-classes to out-source common methods/properties definitely sounds like the best way to go! Is that still the ideal way...

> Using mixin sub-classes to out-source common methods/properties definitely sounds like the best way to go! seems that isn't possible as `self` needs to be `self._data`

Over multiple PRs, my goal is to remove `inherit_names` to make static type checkers aware of those methods. The alternative is to keep `inherit_names` but add `method_name: type_annotation`. Writing those...

I think I addressed all comments but now we also have some small implementation changes. If you could please have another look @jbrockmendel @mroeschke

Are you able to share a SAS file to reproduce this issue?

When I read the entire file (without chunksize), I can re-producde the segfault on master. With chunksize, I get a `UnicodeDecodeError` > File "pandas/pandas/core/strings/accessor.py", line 1795, in > f =...

> I haven't looked too deep but it sounds like the bottleneck is in `darglint/parse/cyk.py`. If cyk.py (and ideally all its imports) are sufficiently type annotated, it might be worth...

> After this PR autotyping will gently nudge people to add type annotations for arguments with a string default value. The suggested type by autotyping might not be correct: an...