###### tags: `Monthly Community Meetings`

# 2023-02-16 Eiffel monthly community meeting: Eiffel event correlation using links

## Attendees

- Emil Bäckmark, Ericsson
- Tobias Persson, Axis Communications
- Mattias Linnér, Ericsson
- Peter Princz, Ericsson
- Magnus Bäck, Axis Communications
- Per-Arne Landström, Ericsson
- Helena Slivova, Tietoevry
- Kristian Sandahl, Linköping University

## Topic
**What makes the Eiffel link concept stand out and is it the only valid solution to correlate events?**
The background is that in the ongoing work to define the [CDEvents](https://cdevents.dev/) protocol, the concept of “links” has not yet been accepted as the way to correlate events. This discussion aims to provide use cases and scenarios where the link model in Eiffel is favorable over for example “span IDs” á la OpenTelemetry.

## Discussion
- Round the table - What are the good things with links in Eiffel? What are the painpoints with links?
- Good
- Semantics
- Bad
- Complicated if you only have a simple usecase
- Links vs Trace/Global/Span IDs and Trace contexts?
- https://opentelemetry.io/docs/concepts/signals/traces/
- https://github.com/honeycombio/buildevents
- https://user-images.githubusercontent.com/361454/57872910-ac9eea00-77c1-11e9-8bdd-db7a870dcd61.png
- Use SBOMs references instead of COMPOSITION links in ArtC events, how does that affect the need for links?
- Theoretically we get the same traceability if we link to SBOMs that, in turn, contain references that we can use to look up other Eiffel entities like artifacts or source changes. However, then you wouldn't have a single Eiffel graph but multiple graphs that need to be tied together via the SBOM references. This may or may not be a problem.
- Would it be possible to convert an SBOM to a composition? Possibly, but would require all entities described in the SBOM to have Eiffel events.
- How do we deal with FOSS components in the SBOM that don't have corresponding events like in-house components do?
- What about temporary baselines? Can there be "preliminary" SBOMs?
- Could be handled just as any preliminary artifact/document probably, but maybe not as part of the SBOM document itself
- Can SBOMs also describe full environments including IUT and Test environment tools/systems?
- We don't know
- Dimensions - Artifacts vs Activities. Are links relevant within both dimensions? Or mostly between them?
- An advantage with having references in the Eiffel events rather than the raw data is that you can have different authorization requirements for the raw data.
- Confidence levels
- This might not be a killer use case for Eiffel event links
- Proposals forward. Eiffel TC to pick and choose :)
1. Initiate an activity to investigate what it would mean to introduce SBOM references in some event(s) and remove the COMPOSITION links and CD events.
- Expensive
1. Could we overlay Eiffel events on top of Open Telemetry “events”?
1. Invite someone from Open Telemetry to present to us and discuss
1. Bring it to the AI community, if our semantic links would be valuable/crucial to enable/simplify AI/ML
1. Bring the question to the Software Center Reporting Workshop