Remove the proto dependency in goldendataset for traces #3322
Conversation
…llector into 3229-span resolve conflit
Co-authored-by: Anthony Mirabella <[email protected]>
|
https://github.com/open-telemetry/opentelemetry-collector/pull/3322/checks?check_run_id=2722314021 failing tests that use the golden data, so seems related to this. |
I think the root cause is that some functions like generateSpanLinks() used to return a nil pointer when the input is nil, but in my previous code, I have to return empty objects since the return type was not a pointer. This results in slightly different generated Spans and Traces than before. I changed the code to match the previous behavior. |
|
I closed this PR by mistake. Then I reopened it and sorry for the inconvience. |
| parts.ArrayVal().Append(pdata.NewAttributeValueString("otelcol")) | ||
| parts.ArrayVal().Append(pdata.NewAttributeValueString("--config=/etc/otel-collector-config.yaml")) | ||
| parts.ArrayVal().Append(pdata.NewAttributeValueString("--mem-ballast-size-mib=683")) | ||
| attrMap["conventions.AttributeProcessCommandLine"] = parts |
There was a problem hiding this comment.
This is a bug, transformed from the constant conventions.AttributeProcessCommandLine to "conventions.AttributeProcessCommandLine". Fixed in #3377.
| if err != nil { | ||
| instrumentationLibrarySpansSlice.Append(*libSpans) | ||
| } |
There was a problem hiding this comment.
This is a bug, fixed in #3377.
There was a problem hiding this comment.
This bug causes actually to always produce empty data, so that is the reason the previous bug was hidden.
Description:
Remove the proto dependency in
goldendatasetfor generating traces. Functions being called by GenerateTraces() will use PData instead of protos for the final constructing of Traces.Link to tracking Issue:
#3229
Testing: The unit test is updated to use PData in test cases.