tagging myself

@scheler, @martinkuba, @MSNev Pl. review.

Do we want this in the spec? Why isn't it defined in the JS SIG's documentation? Is this applicable to more than one language?

Do we want this in the spec? Why isn't it defined in the JS SIG's documentation? Is this applicable to more than one language?

Yes, that is what the RUM SIG desires. IMO, this is schema, and putting it inside a specific language repo doesn't help discoverability much. Also, as part of RUM we intend to define other events for clients in general. Those schemas would anyway end up in the spec (or another common location) repo. Keeping everything organized here would help drive consistency across client schemas.

Do we want this in the spec? Why isn't it defined in the JS SIG's documentation? Is this applicable to more than one language?

In addition to what Ram said, it will also help us in leveraging the tools built for semantic conventions. These are semantic conventions and we already have browser and mobile specific conventions in the spec repo.

@ramthi a few things -

1. These semantic conventions have to first go into the top level semantic_conventions folder as yaml files and then use the otel/semconvgen tool to generate the markdown text that goes into specification//semantic_conventions folder.
2. The attributes you identified as common as Resource level attributes so they go under semantic_conventions/resource/browser.yaml
3. The others go into either semantic_conventions/trace or semantic_conventions/logs depending on whether we plan to send those attributes in Spans (including Span Events) or Events. For the attributes that can go in either, I think you can put them in one and reference from the other. See logs/log-exception.yaml for an example.

@ramthi Another thing to note is that these are only semantic conventions and not strictly schema defining the shape of Events. That is out of scope for both OpenTelemetry semantic conventions and OpenTelemetry schemas. If we want the schema to define the shape of Events, then that will be a new feature addition to the spec.

I am concerned about the following:

• (In)ability of spec maintainers to meaningfully review the proposed semantic conventions. The conventions are pretty deep in the browser territory and reviewers here don't necessarily have the required experience.
• Slower pace of creating such semantic conventions. Everything is slower in spec repo (by design - we require more rigour and scrutiny).
• Unclear if the proposed conventions need to work across more than just JS. Typically spec is concerned about cross-cutting matters.

@jsuereth what do you think?

@tigrannajaryan RUM is not just browser apps monitoring, it includes mobile apps as well and potentially IoT in the future. The mobile apps use opentelemetry-java and opentelemetry-swift. It will be good to have the conventions for all of these in one place.

This PR was marked stale due to lack of activity. It will be closed in 7 days.

This PR was marked stale due to lack of activity. It will be closed in 7 days.

@jsuereth I think this is a topic for Semantic Convention WG to make a call on.

This PR was marked stale due to lack of activity. It will be closed in 7 days.

Closed as inactive. Feel free to reopen if this PR is still being worked on.