[blaipr]

What

Ways for a reader to ask for less, all on existing routes:

• groups=all|matching|none on GET /sources/{id}/dataset (sources and views).
  • matching: only the groups of the returned hosts, member lists cut to them, plus the parents of kept groups (walked up), children cut to kept groups. Group vars untouched (minus exclude_fields).
  • none: no groups key.
  • all stays the default: nothing changes for existing consumers.
• Dotted paths in fields, exclude_fields and filter keys — dataset route, views and builtin endpoints (csv columns too): fields=monitor.name, filter=monitor.status=DOWN.
  • A top-level key of the exact name still wins, so old requests select what they did.
  • Paths walk objects only, never lists.
• host=, group=, filter= on GET /sources/{id}/hosts (sources and views), selecting exactly as /dataset does; names only. Malformed filter = 400.

Why

A narrow dataset read carried every group with its full member list. On a source of ~1,000 hosts and 40 groups:

One host	Raw	gzip
groups=all (default)	444,420 B	24,892 B
groups=matching	3,516 B	709 B
groups=none	2,489 B	613 B

And fields could only pick top-level keys, so a source keeping everything under one key couldn't be trimmed at all (3,047,870 B → 100,303 B with fields=<object>.name&groups=none).

How

• domain/var_path.rs: lookup (filter) and VarSelect (fields/exclude_fields), shared by every route.
• domain/group_trim.rs: the matching selection, generic over owned and borrowed names.
• GroupsMode added to the OpenAPI schemas; groups is part of the filtered ETag.

Tests

• Unit: dotted lookup/keep/drop (literal key wins, lists not walked, keep-then-drop), group trimming (parents, grandparents, empty page), endpoint fields + csv path columns.
• Integration: groups=matching|none|<bad>, default unchanged, dotted fields/exclude (group vars too), dotted filter, /hosts selectors and 400.
• cargo fmt --check, cargo clippy --all-targets -- -D warnings, cargo test: 700 passed.