I also found this confusing. I think it's fair to say the app can't change this.
See https://opentelemetry.io/docs/specs/semconv/http/http-spans/#status with a similar decision (that motivated me to raise this).
The examples of instrumentation library are things like attachable gRPC or mongodb instrumentation. I think this statement describes our client libs, so I think when we look at the spec we should think of ourselves as Instrumentation Library.
Instrumented Library and Instrumentation Library may be the same library if it has built-in OpenTelemetry instrumentation.
The definition of "Ok":
Ok
The operation has been validated by an Application developer or Operator to have completed successfully... When span status is set to Ok it SHOULD be considered final
I think most spans are 'unset' (treated as success downstream) and 'ok' is just used to override 'error'.
I don't really see it as a big deal outside of interpreting the spec. The only 2 practical problems I can imagine are:
- some downstream processor extension (within tracing-opentelemetry or later) might find it more difficult to label this span an "error" since OK is "final".
- some downstream dashboard might put some green check mark next to these that you won't find next to the 'unset' HTTP spans
Originally posted by @westarle in #4937 (comment)
I also found this confusing. I think it's fair to say the app can't change this.
See https://opentelemetry.io/docs/specs/semconv/http/http-spans/#status with a similar decision (that motivated me to raise this).
The examples of instrumentation library are things like attachable gRPC or mongodb instrumentation. I think this statement describes our client libs, so I think when we look at the spec we should think of ourselves as Instrumentation Library.
The definition of "Ok":
I think most spans are 'unset' (treated as success downstream) and 'ok' is just used to override 'error'.
I don't really see it as a big deal outside of interpreting the spec. The only 2 practical problems I can imagine are:
Originally posted by @westarle in #4937 (comment)