Skip to content

Define handler as an async generator in mlrun_footer#197

Merged
rokatyy merged 3 commits intonuclio:masterfrom
gtopper:ML-11876
Jan 28, 2026
Merged

Define handler as an async generator in mlrun_footer#197
rokatyy merged 3 commits intonuclio:masterfrom
gtopper:ML-11876

Conversation

@gtopper
Copy link
Contributor

@gtopper gtopper commented Jan 27, 2026

ML-11876

Required for mlrun/mlrun#9248.

This is necessary because nuclio's Python wrapper performs static checks on the handler, meaning that it cannot simply act as a pass-through for whaterver context.mlrun_handler(context, event) returns.

@gtopper gtopper requested a review from rokatyy January 28, 2026 04:23
@rokatyy rokatyy merged commit 9a291d7 into nuclio:master Jan 28, 2026
2 checks passed
gtopper added a commit to gtopper/mlrun that referenced this pull request Jan 29, 2026
gtopper added a commit to mlrun/mlrun that referenced this pull request Jan 29, 2026
Adds support for streaming responses in serving graphs, enabling
real-time chunk-by-chunk HTTP responses (e.g., for LLM token streaming).

Key changes:
* New `set_streaming(enabled=True)` API on serving functions
* Async streaming handler that yields results as they're produced by
graph steps
* Graph steps can now use generators to stream multiple chunks
* Updated nuclio handler for generator return type support

Using the new functionality introduced in storey 1.11.8 /
mlrun/storey#605.

[ML-11876](https://iguazio.atlassian.net/browse/ML-11876)

Depends on nuclio/nuclio-jupyter#197.

[ML-11876]:
https://iguazio.atlassian.net/browse/ML-11876?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants