Hi Ragas team 👋 — I'm Hassieb from the Langfuse team.
Langfuse Cloud's Fast Preview requires Python SDK v4.7.0+ for real-time data. Ragas' tracing extra currently allows any langfuse>=3.2.4, while parts of the docs still use removed v2 context imports.
From the current integration, the concrete migration looks like:
- Constrain the extra to
langfuse>=4.7,<5.
- Update examples from
langfuse.decorators/langfuse_context to top-level observe and get_client().
- Replace
langfuse_context.score_current_trace(...) with the v4 client scoring API, and use propagate_attributes() for user/session context.
- If
LangfuseTrace.sync_trace() and filter() remain public, make them fetch/filter real v4 API data rather than returning placeholder objects.
- Add a test that imports the real v4 package and runs the documented tracing flow.
The relevant guide is Python SDK v3→v4.
If you need support with the migration, we're happy to jump on a call or help review a PR.
Hi Ragas team 👋 — I'm Hassieb from the Langfuse team.
Langfuse Cloud's Fast Preview requires Python SDK v4.7.0+ for real-time data. Ragas'
tracingextra currently allows anylangfuse>=3.2.4, while parts of the docs still use removed v2 context imports.From the current integration, the concrete migration looks like:
langfuse>=4.7,<5.langfuse.decorators/langfuse_contextto top-levelobserveandget_client().langfuse_context.score_current_trace(...)with the v4 client scoring API, and usepropagate_attributes()for user/session context.LangfuseTrace.sync_trace()andfilter()remain public, make them fetch/filter real v4 API data rather than returning placeholder objects.The relevant guide is Python SDK v3→v4.
If you need support with the migration, we're happy to jump on a call or help review a PR.