[Log Explorer] Group uncategorized data streams by dataset

[weltenwort]

📓 Summary

Currently the list of "uncategorized" data streams in the dataset selector lists all data streams of the same type and dataset separately for each namespace. In order to provide a consistent experience with the integration-owned data streams, we want to only show one entry for each dataset and leave the namespace filtering up to the filter control bar.

• 🔗 related to: [Log Explorer] Add filter controls customization point #158561

✔️ Acceptance criteria

• For a given combination of type and dataset, only one "uncategorized" data stream entry is shown, which uses * for the namespace.
• Selecting the entry of an "uncategorized" data stream creates a data view targeting ${type}-${dataset}-*.
• The namespace filter control offers the known values for the namespace for filtering if this is implemented after [Log Explorer] Add filter controls customization point #158561 is merged.

🎨 Examples

Currently the same uncategorized dataset example in three different namespaces one, two, and thre would be listed as:

• logs-example-one
• logs-example-two
• logs-example-three

Instead we want them to be listed as

• logs-example-*

💡 Implementation hints

• Since the fleet API should not be tied too closely to our UX concerns, we probably want to perform this grouping on the browser side after fetching the list of uncategorized data streams.

[elasticmachine]

Pinging @elastic/infra-monitoring-ui (Team:Infra Monitoring UI)

[mohamedhamed-ahmed]

qq: Do we want to list them as logs-example-* or just logs-example

[weltenwort]

Good question. I used logs-example-* in the ACs because it reflects better what's going on in the query. If we just removed the namespace I assume it would be harder for the user to understand the connection to their data streams.

While we generally try to not leak unnecessary Elasticsearch concepts, I see the uncategorized data streams as an intentional bridge between the ES world and the integration-based world.

Does that seem like a reasonable argument?

[mohamedhamed-ahmed]

yes makes sense to me, at the beginning I thought it would better be consistent with the categorized dataSets, but your argument makes total sense 👍