Update status serialization for Zipkin (was: Remove mention of gRPC status codes) #1186
Conversation
`Span.Status` is not related to gRPC codes anymore
| @@ -138,7 +138,8 @@ TBD: add examples | |||
|
|
|||
| ### Status | |||
|
|
|||
| Span `Status` MUST be reported as a key-value pair in `tags` to Zipkin. | |||
| Span `Status` MUST be reported as a key-value pair in `tags` to Zipkin, unless it is `UNSET`. | |||
| In the latter case it MUST NOT be reported. | |||
There was a problem hiding this comment.
I think we still want to report a non-empty message even if the code is UNSET. But someone knowledgeable about Zipkin should probably review.
There was a problem hiding this comment.
No reason to do that, unnecessary bytes on the wire.
Co-authored-by: Christian Neumüller <[email protected]>
Oberon00
changed the title
|
Since this now is more than editorial, we might want an issue for that, so it can be triaged. I don't think splitting the non-editorial changes off makes much sense, at least in the current master branch state it should be obvious that the spec is outdated here 😃 |
|
@open-telemetry/technical-committee should I do anything else with this or can it be merged? |
LGTM. Will be merged after the compulsory 2 day period (unless we need it more urgently for some reason). |
…tatus codes) (open-telemetry#1186) Co-authored-by: Christian Neumüller <[email protected]>
Span.Statusis not related to gRPC codes anymore.