The OpenTelemetry project is trying to standardize...

# server

[Gregor Zeitlinger]

The OpenTelemetry project is trying to standardize on a context propagation mechanism - and I've found https://github.com/Kotlin/kotlinx.coroutines/issues/284 and would like to know more about it

[elizarov]

What exactly you’d like to know?

[Gregor Zeitlinger]

How can OpenTelemetry use the coroutine context to propagate information (the active span context in this case)?

[Gregor Zeitlinger]

Ideally this could be done in a way that does not introduce any direct dependency to kotlin

[Gregor Zeitlinger]

in go, this is possible, because there is a standard for this: https://golang.org/pkg/context/

[Gregor Zeitlinger]

in java, this comes close: https://grpc.github.io/grpc-java/javadoc/io/grpc/Context.html

[Gregor Zeitlinger]

@elizarov

[elizarov]

The way it is done with all the other thread-based context propagation mechanism (like logging) is this:
• There is some thread-local-based context mechanim that is implemented in the library. The best is that when the context is immutable (like in GRPC) and can be updated only explictly.
• There is a separate “coroutines integration module” that depends on kotlinx.coroutines and provides an intergration element for coroutines context.

[Gregor Zeitlinger]

great - just checked out the repo
I found the MDC propagation you mentioned but not the GRPC module - is this external?

[elizarov]

We don't have GRPC module. Maybe there is 3rd party one. IDK.

[Gregor Zeitlinger]

ok

[elizarov]

It does not have to be in kotlinx.coroutines. We only maintain integration modules for very popular and widely-used libraries, trying to stay away from domain-specific ones.

[Gregor Zeitlinger]

makes sense - we're trying to establish a jvm standard 😎