PyCharm debugger crash

[nhoover]

Works fine when not debugging, but with or without breakpoints, the debugger crashes:

File "C:\Users\nathan\Desktop\nstrava\modules\caching_bar.py", line 37, in bar_graph_data
monthly_distance = [a.groupby([a.index.year, a.index.month]).agg({DISTANCE_HEADER:
File "C:\Users\nathan\Desktop\nstrava\modules\caching_bar.py", line 37, in
monthly_distance = [a.groupby([a.index.year, a.index.month]).agg({DISTANCE_HEADER:
File "C:\Users\nathan\Desktop\Nathan\venvs\nstrava3.10\lib\site-packages\pandas\core\frame.py", line 8413, in unstack
result = unstack(self, level, fill_value)
File "C:\Users\nathan\Desktop\Nathan\venvs\nstrava3.10\lib\site-packages\pandas\core\reshape\reshape.py", line 478, in unstack
return _unstack_frame(obj, level, fill_value=fill_value)
File "C:\Users\nathan\Desktop\Nathan\venvs\nstrava3.10\lib\site-packages\pandas\core\reshape\reshape.py", line 505, in _unstack_frame
unstacker = _Unstacker(obj.index, level=level, constructor=obj._constructor)
File "C:\Users\nathan\Desktop\Nathan\venvs\nstrava3.10\lib\site-packages\pandas\core\reshape\reshape.py", line 133, in init
if num_cells > np.iinfo(np.int32).max:
File "C:\Users\nathan\Desktop\Nathan\venvs\nstrava3.10\lib\site-packages\numpy\core\getlimits.py", line 651, in init
self.dtype = numeric.dtype(type(int_type))
TypeError: 'NoneType' object is not callable

Code is here: https://github.com/nhoover/nstrava/blob/main/modules/caching_bar.py#L37

   # Aggregate gear (bikes and unis/shoes), distance & climbing by year and activities by month
    yearly_distance = [a.groupby([YEAR_COLUMN]).agg({DISTANCE_HEADER: ['sum', 'count']}) for a in acts]
    yearly_climbing = [a.groupby([YEAR_COLUMN]).agg({CLIMB_HEADER: ['sum']}) for a in acts]
    monthly_counts = [a.groupby([a.index.year, a.index.month]).size() for a in acts]
    monthly_distance = [a.groupby([a.index.year, a.index.month]).agg({DISTANCE_HEADER:
                                                                      ['sum']}).unstack(fill_value=0) for a in acts]
    monthly_climbing = [a.groupby([a.index.year, a.index.month]).agg({CLIMB_HEADER:
                                                                      ['sum']}).unstack(fill_value=0) for a in acts]

[nhoover]

It happens for lots of other people too, in python 3.10 https://youtrack.jetbrains.com/issue/PY-52137

[nhoover]

This code reproduces the crash

import pandas
df = pandas.DataFrame({'A': [1, 2], 'B': [0.5, 0.75]})
rec = df.to_records()
print(rec)

[nhoover]

The root cause of this issue is in numpy. It should be fixed in the next release, 1.22.4: numpy/numpy#21008

[nhoover]

Here's the release page where numpy 1.22.4 will show up https://github.com/numpy/numpy/releases

[nhoover]

Looks like they finally fixed the underlaying issue in cython, although it may be a while until it propagates into an actual numpy fix. cython/cython#4735

[nhoover]

Bummer - numpy 1.22.4 was just released and the notes say it should fix all these debugging problems everyone is having. But it doesn't for me.

[nhoover]

Tried installing cloning the pandas repo and then installing it's setup.py in my nstrava venv. Had to install cython to build and it takes a really long time then fails with

Download error on https://pypi.org/simple/: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:997) -- Some packages may not be found!
No local packages or working download links found for pandas==1.5.0.dev0+809.gd2a7eff002
error: Could not find suitable distribution for Requirement.parse('pandas==1.5.0.dev0+809.gd2a7eff002')

[nhoover]

Problem is that both pandas and numpy need to be rebuilt with the new and fixed cython to fix this.

[nhoover]

Supposedly pandas 1.4.3 will fix this and may finally be released this weekend or soon. Don't hold the breath though as it was initially scheduled for release 4/30/2022. pandas-dev/pandas#46610

[nhoover]

Finally fixed by upgrading to pandas 1.4.3 and numpy 1.23.0 in build v626 6/24/2022 - debugging works!