to_dataframe() causes Colab kernel crash when processing BQ INTERVAL type

[wlongabaugh]

Environment details

• OS type and version: Google Colab notebook (created 11/17/22)
• Python version: "Python 3.7.15"
• pip version: "pip 21.1.3 from /usr/local/lib/python3.7/dist-packages/pip (python 3.7)"
• google-cloud-bigquery version: "Version: 3.3.6"

Steps to reproduce

1. Run following 4-line snippet on a Google Colab notebook, replacing project ID
2. Kernel crashes. This did not happen a week ago on a notebook that put an TIMESTAMP INTERVAL in a dataframe; it was working for months up until this week. This code snippet is the smallest I can create to show the problem.
3. Did not report to Colab issue tracker, as this seems likely to be an underlying library problem, perhaps tied to the 3.3.6 release 13 days ago?
4. I did check that INTERVAL has been a feature request in the past, perhaps incomplete? But the code has worked fine up until recently.

Code example

from google.colab import auth
from google.cloud import bigquery

auth.authenticate_user() 
bigquery.Client('<your-project-id>').query("SELECT MAKE_INTERVAL(second => 1) as duration").to_dataframe()

Stack trace

Google Colab notebook kernel crash reported: "Your session crashed for an unknown reason". Logs are uninformative.

[wlongabaugh]

Looking at Colab release notes from last Friday it looks like they upgraded google-cloud-bigquery from 1.21.0 to 3.3.5, which I guess explains the problem. I guess this just makes this a duplicate of "Support INTERVAL data type" #826. I will close this issue.