kubeflow / kubeflow/mcp-server
Bump kubeflow dependency from 0.4.1 to 0.5.0 for milestone 0.2
- Dominant language
- Python
- Stars
- 44
- Forks
- 54
- Avg merge
- 2d 18h
- Merged PRs (30d)
- 29
Description
## Summary
SDK 0.5.0 is additive for Trainer, no `TrainerClient` API removals. The bump unblocks Spark batch jobs (KEP-107), aligns Katib/optimizer work (#105) with `OptimizerClient`, and ships `kubeflow-trainer-api` 2.3.0.
## Motivation
- [SDK 0.5.0 release](https://github.com/kubeflow/sdk/releases/tag/0.5.0) : Spark `submit_job`, `PipelinesClient`, `Runtime.kind`, Trainer API 2.3.0
- Kubeflow MCP milsetone 0.2 scope: Katib (#34 / #105), Spark (#5), Hub (#181)
- Spark batch lifecycle requires SDK ≥0.5.0
## Dependencies
- #105
- #5
- #181
- 0.1.2 patch release - Out of scope (stays on Kubeflow SDK 0.4.1)
## References
- [CHANGELOG-0.5.md](https://github.com/kubeflow/sdk/blob/main/CHANGELOG/CHANGELOG-0.5.md)
Contributor guide
Assessment
This issue has not been assessed yet.