[BEAM-10764] Make is_in_ipython robust

• Type: Improvement
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Fix Version/s: Missing
• Component/s: runner-py-interactive
• Assignee: Ning
• Reporter: Ning
• Created: 19/Aug/20 18:58
• Updated: 13/Jan/22 21:48
• Resolved: 15/May/21 18:01

Description

`is_in_ipython` determines if current code execution is within an IPython environment by attempting to fetch an IPython kernel through `IPython.get_ipython()`.

If IPython dependency is not available or a `None` is fetched, the result would be False.

We've been seeing some users using corrupted IPython dependency in their code base.
If an IPython dependency is present but throws a non ImportError exception, it will break the Beam usage.
I assume the similar errors would happen if the user uses an IPython dependency outside the range of versions in setup.py.

I decide to make the function best effort so that it always returns False when errors occur.

Issue Links

• links to: GitHub Pull Request #12641