Codex uses and ships with bubblewrap on Linux and will attempt to use the version installed on the path before falling back to the shipped version with a warning message.
You should be able to configure the sandbox using https://developers.openai.com/codex/agent-approvals-security if you are a person who prefers the convenience of codex being able to open the sandbox over an externally enforced sandbox like jai.
I wasn't familiar with the PIP acronym so I asked $AI:
> A PIP (Performance Improvement Plan) is a formal document a company uses when they believe an employee isn’t meeting expectations. It’s framed as “support,” but depending on the environment, it can be anything from a genuine improvement tool to a pre-termination protocol.
Eclipse Collections also comes with an API that seems nicer than the one provided by the standard library - so it might be worth checking out even if you're not interested in primitives.
I will make any excuse to use Streams but understand the negativity. They are difficult to debug and I feel the support for parallelism complicated, and in some cases even crippled, the API for many common use cases.
> I gotta say, I'm surprised that this needed to be explained to the manager. If someone asks a worker whether they want to do something and the worker say "no", the manager's instinct is to panic and escalate before trying to clarify anything?
It's possible those attempts were left out of the story for narrative reasons - it is, like you said, a good example for the article. But let's pretend they weren't and that is how it actually went down: The manager is human and will not navigate every given situation optimally. Hopefully organizations will become increasingly aware of this over time and continuously improve the methodology for selecting and training future managers.
reproduced: 97.02% good: 17586 bad: 511 fail: 30 unknown: 0
This, statistics for other architectures, and the reasons for unreproducibility can be found at https://reproduce.debian.net.