fix: gate config no-op skip on a populated flag, not connection state - #32
Conversation
… fetch add_field/prefer_typed no longer gate their no-op skip on connection/topic state (_record_is_live). Instead, TeslemetryStreamVehicle tracks whether its record has ever been populated (by a push event or a REST fetch); an unpopulated vehicle awaits a single-flight get_config() before deciding, a populated one trusts the optimistically-updated record outright. A disconnect notification unpopulates the record so a call landing in the reconnect window (before the next connection's config snapshot arrives) refetches instead of trusting stale pre-disconnect data. This replaces the mechanisms proposed in #30 (a second dispatch pass for listeners added mid-event) and #31 (per-connection snapshot tracking) with a smaller, self-contained fix scoped to the vehicle itself.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a5fc9a777f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if self._populated: | ||
| return |
There was a problem hiding this comment.
When TeslemetryStream is constructed with an explicit topic allowlist that omits config (for example, only vehicle_data), the first REST fetch sets _populated and this early return remains effective for the entire connection. Because stream.py sends that allowlist unchanged, the internal listener cannot observe subsequent config changes from another client, so add_field or prefer_typed can silently skip a required PATCH using stale state. Preserve topic-awareness in this gate or re-fetch when config events are not subscribed.
AGENTS.md reference: AGENTS.md:L18-L18
Useful? React with 👍 / 👎.
| if req.status == 404: | ||
| # No config exists for this vehicle yet - an authoritative | ||
| # answer (empty), not a missing one. | ||
| self._populated = True |
There was a problem hiding this comment.
If this vehicle has cached fields or preferTyped, is disconnected, and the lazy config GET then returns 404 because its server-side config was removed, this marks the old record populated without clearing it. The following no-op check can therefore skip recreating a missing field or restoring prefer_typed based on stale pre-disconnect values; reset both cached pieces to the authoritative empty/default state before setting _populated.
AGENTS.md reference: AGENTS.md:L21-L21
Useful? React with 👍 / 👎.
1 participant
Intent
add_field/prefer_typed's no-op skip trustedfields/preferTypedbased on connection/topic state (_record_is_live()), not on whether the record had ever actually been confirmed against the server - leaving two races: a vehicle discovered mid-dispatch of its own config event never sees that event and sits unseeded, and a field-config call landing betweenconnect()and that connection's config snapshot can match against a stale pre-disconnect record.TeslemetryStreamVehicletracks_populated, set by a successfulget_config()(200 or 404 - both are an authoritative answer) and by every_on_config_eventpush.add_field/prefer_typedawait_ensure_populated()first - a no-op if already populated, otherwise a single-flightget_config()fetch that concurrent callers (e.g. a batch oflisten_*calls at HA integration setup) join instead of each starting their own GET._on_connection_eventlistener (registered viaasync_add_connection_listenerat construction, alongside the existing config-sync listener) unpopulates the record on disconnect, so a call landing in the reconnect window re-fetches instead of trusting stale data._record_is_live()and its connection/topic coupling are gone entirely - optimistic updates from push events, plus the lazy fetch for the unpopulated gap, are the whole mechanism now.tests/test_stream_lifecycle.pyadapts fix: seed a vehicle discovered mid-dispatch from its triggering config event #30's scenario: the mid-dispatch-discovered vehicle stays unpopulated (no seeding trick needed) and self-corrects via the lazy fetch on its first field-config call.tests/test_reconnect_config_window.pyadapts fix: close reconnect-window gap in add_field/prefer_typed no-op skip #31's scenario against the new design.tests/test_config_listener_lifecycle.pycovers the populated/unpopulated gating directly.