fix(db-manager): address review findings on the schema diagram

- The auto-select effect read the state it wrote, so unchecking the last
  table refilled the schema and an empty schema looped it until the pane
  died. It now runs once per entry into the mode.
- Returning to the diagram re-queued the foreign-key query every time; it
  is now re-read only when the schema itself was reloaded.
- A switch to a non-PostgreSQL database left the diagram on screen with no
  toggle to leave it by.
- A checkbox click in diagram mode also queued the per-table foreign-key
  job of the data view, whose result the diagram never reads.
- Update the registered-counter tally in docs/feature-telemetry.md.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MXECAKye6KgN13wznsuRzP
This commit is contained in:
Diego Imbert
2026-09-08 18:37:50 +02:00
co-authored by Claude Opus 5
parent d2842e0751
commit 520bfb7abb
3 changed files with 91 additions and 21 deletions
+4 -4
View File
@@ -4,10 +4,10 @@
anonymous usage-stats payload. It answers "does anyone use this, and which variant do they pick"
without any identifying data leaving the instance.
It currently carries 42 registered actions across seventeen features (`ai_session`, `ai_chat`,
`ai_fix`, `ai_agent`, `ai_agent_eval`, `app_sandbox`, `datatable`, `flow_editor`, `flow_run`,
`flow_step`, `run_form`, `debugger`, `trigger`, `command_script`, `hub_script`, `usage_meter`,
`sso_groups_claim`). Nearly all of the
It currently carries 43 registered actions across eighteen features (`ai_session`, `ai_chat`,
`ai_fix`, `ai_agent`, `ai_agent_eval`, `app_sandbox`, `datatable`, `db_manager`, `flow_editor`,
`flow_run`, `flow_step`, `run_form`, `debugger`, `trigger`, `command_script`, `hub_script`,
`usage_meter`, `sso_groups_claim`). Nearly all of the
product is uninstrumented, so new user-facing work is the opportunity to change that.
## When to instrument