feat(runtime): cli, telemetry (zenoh), fault tolerance#119
Conversation
maxxgx
force-pushed
the
max/runtime-cli-telemetry
branch
from
6c1c3e3 to
57d3126
Compare
maxxgx
force-pushed
the
max/runtime-cli-telemetry
branch
from
57d3126 to
9b6567d
Compare
There was a problem hiding this comment.
To follow the same CLI pipeline with the studio, we have merged this PR
https://github.com/open-edge-platform/physical-ai-studio/pull/569/changes
Next, we need to move these changes to physicalai here. We could then create a jsonargparse based CLI that can mimic the exact behavior of that of studio's and that can be extendible for studio as well.
There was a problem hiding this comment.
is this observer subpackage for telemetry only ?
There was a problem hiding this comment.
I was thinking we could handle telemetry with a Callback mechanism. Thoughts ?
There was a problem hiding this comment.
I think we should have both callbacks and a pub-sub telemetry mechanism, as they target different use cases.
I'd imagine callbacks are more simpler and more convenient for someone using our library in their own project. While zenoh pub-sub is useful for accessing telemetry/events from external processes, for example, using the standalone runtime for a demo with visualization
| @classmethod | ||
| def from_config(cls, config: RuntimeConfig) -> PolicyRuntime: |
There was a problem hiding this comment.
again, the following pr can massively simplify the logic here
https://github.com/open-edge-platform/physical-ai-studio/pull/569/changes
2 participants
Summary
Why
Validation
Breaking changes
Related issues