The official changelog calls it a "major release," but from where I sit, it looks more like a breaking change they forgot to announce properly. My team's CI pipelines started failing immediately after the 0.1.0 upgrade, and the error messages are less than helpful.
I'm talking about two specific, disruptive issues:
* The `tracing` module seems to have been restructured. Our previous imports, like `from langsmith.run_helpers import traceable`, now throw `ModuleNotFoundError`. The documentation hasn't caught up, so we're left guessing the new correct import path.
* Client initialization is now brittle. A simple `Client()` instantiation, which worked fine with environment variables before, is now throwing authentication errors in some of our services. The fallback logic appears to be broken.
This isn't just an annoyance. It's a vendor management red flag. When a core SDK update breaks basic functionality without clear, actionable migration notes, it forces costly workarounds and erodes trust. It feels like we're paying to be beta testers.
Has anyone else hit these specific problems? More importantly, has anyone found a reliable workaround or the actual updated import paths? I need to decide if I'm rolling back our deployment or spending the day reverse-engineering their package.
—Daniel
Trust but verify.
Yeah, that's a rough start. I've seen a few other reports about the import path. The `traceable` decorator moved to `langsmith.trace`. You can try `from langsmith import trace` and then use `@trace.traceable` as a direct workaround for now.
It's a valid point about the communication. A major release bump, even to 0.1.0, should come with a migration guide, not just a changelog. When core features like environment variable fallback break, it pushes the debugging burden onto every integration team. It might be worth raising this directly on their GitHub issues - sometimes that gets a faster documentation update than waiting for the official docs to refresh.
Keep it real, keep it kind.