Skip to content

Conversation

@ravwojdyla
Copy link
Contributor

No description provided.

@ravwojdyla
Copy link
Contributor Author

This way FlinkRunner ends up in META-INF/services:

➜  scio git:(flink) ✗ cat META-INF/services/com.google.cloud.dataflow.sdk.runners.PipelineRunnerRegistrar
com.dataartisans.flink.dataflow.FlinkRunnerRegistrar$Runner
com.google.cloud.dataflow.sdk.runners.DataflowPipelineRegistrar$Runner
com.google.cloud.dataflow.sdk.runners.DirectPipelineRegistrar$Runner

gets picked up by Dataflow sdk, and thus can be used if you depend on flink-dataflow.

@mxm
Copy link
Contributor

mxm commented Mar 11, 2016

Thanks for the PR. Could you explain what you use this for?

@ravwojdyla
Copy link
Contributor Author

This is the mechanism of discovering Dataflow runners please read:

Spark runner is doing something similar (here) but they hard code their runner, instead of generate them at compile time ...

@ravwojdyla
Copy link
Contributor Author

Will submit PR to beam repo too.

@mxm
Copy link
Contributor

mxm commented Mar 12, 2016

Thanks for the links. I see that this can be useful. Going to merge in a bit.

@mxm
Copy link
Contributor

mxm commented Mar 15, 2016

Merged in a1e1c74.

@mxm mxm closed this Mar 15, 2016
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