Alexander Wert

Option 2b is cleaner than 2a in my opinion. 2a is very misleading as indicates that it's a single attribute but actually is resolved to a sub-namespace of multiple attributes....

`elasticsearch/elasticsearch-rest-5.0` and `elasticsearch/elasticsearch-rest-6.4` already have Java tests. Moved tests to Java for `elasticsearch/elasticsearch-rest-7.0` as part of this open PR: - open-telemetry/opentelemetry-java-instrumentation#8799.

/cc @jlvoiseux

In SemConv we differentiate between `.` and `_`, in code generation usually not. So I'd say conceptually it's an issue with code gen that needs a cleaner long-term solution. IMHO,...

To make CI happy we need also: https://github.com/open-telemetry/build-tools/pull/316

> One question though - should we enforce then a presence of this property for registry? yeah, I guess we could extend https://github.com/open-telemetry/build-tools/pull/316 to include it in the validity checks

> Could you please create an issue to update schema definition (or just update the schema definition) in the https://github.com/open-telemetry/build-tools/blob/main/semantic-conventions/semconv.schema.json to include it? Done, see: https://github.com/open-telemetry/build-tools/pull/317

> Since the weaver PR merged, you should be able to actually update the weaver `attribute_group.md.j2` template to _use_ this value in this PR. @jsuereth Done with the last commit.

> @AlexanderWert I think you need to run `make table-generation` to re-generate registry MD files with display name. That's not working yet, I'm getting a validation error from weaver. So,...

> The only thing is: Should we mark the new field as required? The Jinja template now assumes it is but if one attribute group misses it the md rendering...