Bug: Numpy version issue causes Jupyter notebook kernel crash when outputting results as Pandas

[prrao87]

Kùzu version

v0.4.2

What operating system are you using?

macOS Sonoma 14.5

What happened?

When I use Python 3.12 and try to run the get_as_df() method in a Jupyter notebook cell, I get the following error where the kernel dies.

On debugging this further with @mxwli, we found that it's related to a numpy version issue.

The moment I run this command pip install numpy~=1.26, the get_as_df() method works.

We should probably find a way to solve this issue without the user having to manually manage their numpy version that's older than current Pandas versions require.

Are there known steps to reproduce?

Run the attached notebook. The CSV file has this schema:

id,name,age
p1,Lisa,25
p2,James,28
p3,Carol,32
p4,David,31

The query run is below:

res = conn.execute(
    """
    LOAD FROM 'data/person.csv' (header = true)
    RETURN id, name, CAST(age, "INT64") AS age
    """
    )
print(res.get_as_df())

test.ipynb.zip