[lxsaah]

Description

Stage 2 of 2 for #390, into the #392 feature branch.

• OutboundLink::on(handle, address) / InboundLink::on(handle, address) start a detached link on a connector registered with AimDbBuilder::connect.
  • A bare address ("sensors/temp") takes the connector's scheme.
  • A full URL is accepted if its scheme matches the connector's, so #[link_address] keys keep working.
• A different scheme, or a handle from another builder, becomes a ConfigError reported by build(). An out-of-range index gets the same message, since only another builder's handle can produce one. The check is connector::resolve_link_url, called at the top of push_outbound and inbound_link, next to the existing link checks. The error names the record and the address as written.
• Both link builders get a connector: Option<ConnectorHandle> field, carried through new, attach and detach. RecordRegistrar gets the builder's id (builder_id).
• Docs: rustdoc on on, ConnectorHandle and connect. The usage guide's "Links as values" example uses connect + on.

Tests

• handle_links_take_the_connector_scheme: with connectors [mqtt, knx], index 1 really yields knx:// in both directions, and a matching full URL is accepted.
• handle_links_reject_wrong_scheme_and_foreign_handles: wrong scheme in both directions, another builder's handle in both directions, and an out-of-range index. Each case gets its exact error and address, and nothing is registered.
• handle_links_build_and_foreign_handles_fail_it: the target shape from [FEAT] AimDbBuilder::connect returns a ConnectorHandle; links through a handle need no scheme #390 through build(). A handle from the same builder builds and subscribes commands/temp/indoor; another builder's handle gives InvalidConfiguration naming temp.indoor.

Checks

CI doesn't run on PRs into the feature branch, so I ran these locally:

• aimdb-core tests with std,tracing,observability (310) and with alloc
• the thumbv8m build
• all nine of core's clippy lines from the Makefile (-D warnings)
• cargo doc with -D warnings
• cargo check -p tokio-mqtt-connector-demo
• the MQTT link_ext_tests
• the aimdb-data-contracts lib tests

The changelog for #390 follows in a separate PR, as for #389.

Related Issue

• Part of [FEAT] AimDbBuilder::connect returns a ConnectorHandle; links through a handle need no scheme #390 / [FEAT] reg.input / reg.output with connection handles, so a wrong wiring is a compile error #388 (PR feat: reg.input / reg.output, connector handles and publish / subscribe (#388) #392)

🤖 Generated with Claude Code