Specify allowed characters for Baggage keys and values #3801
Conversation
Signed-off-by: Yuri Shkuro <github@ysh.us>
Co-authored-by: Reiley Yang <reyang@microsoft.com>
|
Note: this could be a breaking change e.g. for Go SDK (open-telemetry/opentelemetry-go#3685), whose current implementation is wrong imo. |
Signed-off-by: Yuri Shkuro <github@ysh.us>
There was a problem hiding this comment.
Note: this could be a breaking change e.g. for Go SDK (open-telemetry/opentelemetry-go#3685), whose current implementation is wrong imo.
I would say that we chose the wrong decision in OTel Go. Still we might try to fix it in some backwards-compatible way (by extending the API).
TL;DR; For me this is not a blocker for this PR.
|
I am working towards making a fix in OTel Go: open-telemetry/opentelemetry-go#4804 |
Co-authored-by: Robert Pająk <pellared@hotmail.com>
|
Java is also not compliant with this. Currently, we only propagate keys/values that meet this definition, but we allow arbitrary keys for in-memory usage. |
Signed-off-by: Yuri Shkuro <github@ysh.us>
|
I support @pellared comment on different concerns (the in-process API and the inter-process propagation). Maybe we can adjust this PR to only mention that invalid keys MUST NOT be propagated? Then we can spin off a separate issue/PR discussing on (Also this would be easy to do as it would align with what some SIGs already do) |
|
This PR was marked stale due to lack of activity. It will be closed in 7 days. |
|
Updated to allow arbitrary strings in the API, but caution that Propagators may not allow some keys. |
Co-authored-by: Robert Pająk <pellared@hotmail.com>
|
cc @open-telemetry/go-maintainers - the new wording makes Go implementation non-compliant, but it's not a breaking change for its API, you will just need to relax the checks (i.e. never return errors from |
|
Let's remind maintainers in the call next Monday about this, but otherwise we are fine to go. |
|
@carlosalberto are we good to merge? |
Changes