Current release guide
Upgrade to opentine 0.5
Version 0.5.0 is an additive interoperability release. It adds OpenTelemetry GenAI export, shell-level trace import, stable JSON output for selected read commands, and live LangChain/LangGraph capture. It does not introduce a storage format change.
Upgrade without migrating stored data
python -m pip install --upgrade "opentine==0.5.0"
python -c "import opentine; print(opentine.__version__)"
# Existing stored data is not rewritten
tine verify representative-v0.4.tine
tine show representative-v0.4.tine --json
tine fsck --repo .Portable *.tine files remain format v2 and .tine/repositories retain the v3 object model. There is no v0.4-to-v0.5 migration command, and opening existing data does not rewrite it.
Stored-data compatibility
The v0.5 compatibility gate loads, reads, and verifies golden portable artifacts and repositories produced by the published v0.3.0 and v0.4.0 packages. Existing v1 behavior is also unchanged: Run.load() migrates a v1 artifact to v2 in memory, and re-saving is the one-way format upgrade.
Install the live framework integration only when needed
# Only when using the live LangChain/LangGraph callback handler
python -m pip install "opentine[langchain]==0.5.0"The new [langchain] extra installs langchain-core for the callback protocol used by both LangChain and LangGraph. It is optional and is not included in [all]. The [all] extra also continues to exclude MCP.
Application compatibility checklist
- Upgrade every process that reads or writes shared opentine data.
- Verify representative portable artifacts and run
fsckon repositories. - Keep using the returned IDs from fork, replay, import, and repository operations.
- Check third-party integration dependency ranges before upgrading their environment.
- Do not batch re-save artifacts; v0.5 requires no data conversion.
Continue with tine import, OpenTelemetry interoperability, or live LangChain and LangGraph capture. The v0.4 guide remains the reference for fork-act identity introduced in that release.