Hi all, I've been poking around the beam source code trying to determine whether it's possible to get the definition of a pipeline via beam's gPRC-based services. It looks like the message types are there for describing a Pipeline <https://nuget.pkg.github.com/apache/beam/blob/master/model/pipeline/src/main/proto/beam_runner_api.proto> but as far as I can tell, they're only used by JobService.Prepare() for submitting a new job.
If I were to create a PR to add support for a JobService.GetPipeline() method, would that be interesting to others? Is it technically feasible? i.e. is the pipeline definition readily available to the job service after the job has been prepared and sent to the runner? Bigger picture, what I'm thinking about is writing a UI that's designed to view and monitor Beam pipelines via the portability abstraction, rather than using the (rather clunky) UIs that come with runners like Flink and Dataflow. My thinking is that using beam's abstractions would future proof the UI by allowing it to work with any portable runner. Right now it's just an idea, so I'd love to know what others think of this. thanks! -chad
