- Update langfuse deserialization to not require
span_handlerkey (#3378)
- Fix langfuse tracer when input messages is
None(#3324)
- [breaking] Migrate to Langfuse Python SDK v4 (#3126)
- Track test coverage for all integrations (#3065)
- Log generation_kwargs in Langfuse spans (#3017)
- (langfuse) Replace MagicMock with Mock to prevent RAM exhaustion (#2543) (#2798)
- Simplify pydoc configs (#2855)
- Remove unused allow-direct-references (#2866)
- Add ANN type annotations to google_genai, hanlp, jina, langfuse, lara (#2990)
- Enforce ruff docstring rules (D102/D103/D205/D209/D213/D417/D419) in integrations 21-30 (#3010)
- (langfuse) Apply trace_id from tracing context to Langfuse traces (#2781)
- Make fmt command more forgiving (#2671)
- [breaking] Langfuse - drop Python 3.9 and use X|Y typing (#2709)
- (langfuse) Add embedder usage metrics for langfuse (#2542)
- Correct token usage accounting, fix trace polling in tests (#2594)
- Remove Readme API CI workflow and configs (#2573)
- (langfuse) Embedder, retriever and generator as obs. type (#2497)
- Enhancement: Adopt PEP 585 type hinting (part 4) (#2527)
- (langfuse) Log levels (#2522)
- Chore: Upgrade langfuse dep, observation types require version>=3.3.1 (#2493)
- Flatten usage_details dict (#2491)
- Change pytest command (#2475)
- Usage_details instead usage (#2481)
- Feat: Langfuse - add support for tool and agent observation types (#2490)
- Langfuse - add py.typed; fix testing with lowest deps (#2458)
- Add pydoc configurations for Docusaurus (#2411)
- Install dependencies in the
testenvironment when testing with lowest direct dependencies and Haystack main (#2418)
- Remove ruff exclude and fix linting in Langfuse integration (#2257)
- Migrate langfuse to v3 (#2247)
- Avoid mixed Langfuse traces in async envs (#2207)
- Make tests successfully run from forks (#2203)
- Standardize readmes - part 2 (#2205)
- Add AmazonBedrockChatGenerator to supported chat generators in Langfuse (#2164)
- Simpler generation spans, use Haystack's to_openai_dict_format (#2044)
- Properly cleanup Langfuse tracing context after pipeline run failures (#1999)
- Pin langfuse<3.0.0 (#1904)
- Align core-integrations Hatch scripts (#1898)
- Update md files for new hatch scripts (#1911)
- Remove black (#1985)
- Use Langfuse local to_openai_dict_format function to serialize messages (#1885)
- Feat: Add detailed tracing for GoogleGenAIChatGenerator (#1887)
- Add Langfuse pins (#1837)
- Test: Update how test skipping works in Langfuse (#1756)
OllamaGeneratorsupport in Langfuse (#1759)
- Add more options to LangfuseConnector (#1657)
- Enhance Langfuse ToolInvoker span naming (#1682)
- Langfuse - remove warning "Creating a new trace without a parent span is not recommended"
- [breaking] Make sure to JSON serialize objects before setting content tags (#1627)
- Adapt Ollama metadata to OpenAI format; support Ollama in Langfuse (#1577)
- Unify traces of sub-pipelines within pipelines with Langfuse (#1624)
- Correctly set pipeline input and output by updating DefaultHandler (#1589)
- Add trace_id to output of LangfuseConnector (#1587)
- Fixing typo in Langfuse API docstring
- Langfuse - wait before retrieving the trace in
test_custom_span_handler(#1359) - Langfuse - make polling more robust in tests (#1375)
- Review testing workflows (#1541)
- Remove Python 3.8 support (#1421)
- Add custom Langfuse span handling support (#1313)
- LangfuseConnector - add httpx.Client init param (#1308)
- End langfuse generation spans properly (#1301)
- Add LangfuseConnector secure key management and serialization (#1287)
- Chore: Langfuse - pin
haystack-ai>=2.9.0and simplify message conversion (#1292)
- Warn if LangfuseTracer initialized without tracing enabled (#1231)
- Use text instead of content for ChatMessage in Llama.cpp, Langfuse and Mistral (#1238)
- Chore: Fix tracing_context_var lint errors (#1220)
- Fix messages conversion to OpenAI format (#1272)
- Add support for ttft (#1161)
- Adopt uv as installer (#1142)
- Fixed TypeError in LangfuseTrace (#1184)
- Update ruff linting scripts and settings (#1105)
- Fix: Add delay to flush the Langfuse traces (#1091)
- Add invocation_context to identify traces (#1089)
- Langfuse - support generation span for more LLMs (#1087)
- Remove usage of deprecated
ChatMessage.to_openai_format(#1001)
- Add link to langfuse in LangfuseConnector (#981)
- Do not retry tests in
hatch run testcommand (#954)
- Retry tests to reduce flakyness (#836)
Langfuse- replace DynamicChatPromptBuilder with ChatPromptBuilder (#925)- Remove all
DynamicChatPromptBuilderreferences in Langfuse integration (#931)
- Ci: install
pytest-rerunfailureswhere needed; add retry config totest-covscript (#845) - Chore: Update Langfuse README to avoid common initialization issues (#952)
- Chore: langfuse - ruff update, don't ruff tests (#992)
- Feat: add support for Azure generators (#815)
- Langfuse integration (#686)
- Performance optimizations and value error when streaming in langfuse (#798)
- Use ChatMessage to_openai_format, update unit tests, pydocs (#725)
- Chore: change the pydoc renderer class (#718)
- Docs: add missing api references (#728)