Repository navigation
feat(stats): restore GitHub history and track FTW Cloud energy usage - #50
Conversation
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
ftw-webapp | 9001b8d | Commit Preview URL Branch Preview URL |
Oct 17 2026, 02:57 PM |
|
Stable-release audit update: the reviewed privacy and retention fixes are now on this PR at head The new head:
Keep this PR in draft. The live stats Worker still runs the old head and was observed hiding the total while exposing low-count version and driver labels. Before merge or production deploy we still need:
The threshold counts reports, not unique boxes; the UI and README now say that directly. |
frahlg
changed the title
frahlg
changed the title
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![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: 9001b8d886
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".
| } else if (estimate && typeof estimate.value === 'number') { | ||
| setText('overview-installations','≈ ' + number(estimate.value)); setText('installation-period',estimate.window_start + ' to ' + estimate.window_end + (estimate.state === 'stale' ? ' · stale source' : '')); | ||
| } | ||
| if (estimate && typeof estimate.value === 'number') setText('installation-method',estimate.reports + ' reports over ' + estimate.days + ' complete days. About one successful report per installation per day. Counts reporting active installations, not people or all-time installs; opt-outs, offline boxes, failed sends, restarts and test boxes affect the estimate. Source: ' + estimate.source + '.'); |
There was a problem hiding this comment.
When both collectors have data, the branch above displays the 24-hour usage.installations value, but this unconditional assignment replaces its explanation with the unrelated seven-day fleet estimate methodology and source. The dashboard can therefore show one installation count while claiming it was derived from a different report set; only render this estimate explanation when the estimate is the value selected, and provide the usage-cell provenance otherwise.
Useful? React with 👍 / 👎.
| if (u.installations && u.installations.state === 'visible') { | ||
| setText('overview-installations', format(u.installations)); setText('installation-label','Reporting installations'); setText('installation-period','Opted in · last 24 closed hours'); | ||
| } else if (estimate && typeof estimate.value === 'number') { | ||
| setText('overview-installations','≈ ' + number(estimate.value)); setText('installation-period',estimate.window_start + ' to ' + estimate.window_end + (estimate.state === 'stale' ? ' · stale source' : '')); |
There was a problem hiding this comment.
If a usage snapshot was saved before the source began returning the documented disabled state, dashboardData retains that snapshot while setting u.state to paused; this condition still selects its historical installation cell and labels it as the current “last 24 closed hours” value, even when a fresh fleet estimate exists. Gate this branch on a connected, sufficiently fresh usage state so a pause cannot present an old count as current.
Useful? React with 👍 / 👎.
| installation_estimate: installationEstimate(feed?.days ?? (fleetView['days'] as {date:string;reports:number|null}[] ?? []), nowMs, cloudFleet?.observed_at ?? freshness['fleet'] ?? null, feed ? 'cloud.ftw.energy' : 'retired relay archive'), | ||
| fleet_current: feed, | ||
| usage: { ...(usage ? JSON.parse(usage.view_json) as Record<string, unknown> : { minimum: 20 }), state: usageState }, |
There was a problem hiding this comment.
When PUBLIC_MIN_REPORTS is configured above 10, the legacy fleet view honors it, but /api/public also returns fleet_current verbatim from a feed whose parser fixes its minimum at 10. Counts and dimension labels between 10 and the configured threshold are consequently exposed through the public API, bypassing the operator-selected privacy floor; omit this raw feed or re-filter it using publicMinimum(env.PUBLIC_MIN_REPORTS).
Useful? React with 👍 / 👎.
1 participant
Update stats.ftw.energy for Core, drivers, Cloud, the native app, web app and website. Restore saved GitHub charts after the repository set grew from three to six. Preserve per-point coverage and break star lines when the set changes, so adding repos does not look like growth. Include beta releases when selecting the latest published release.
Use the public Cloud daily-report feed to estimate active reporting installations: 156 reports over the seven complete UTC days 10–16 October gives about 22. This is not unique people or all-time installations. Missing/withheld days hide the estimate; opt-outs, offline boxes, failed sends, restarts and test boxes can bias it. Replace overlapping imported relay history rather than adding it. Keep the retired relay archive clearly labelled.
Add an hourly, bounded collector for public Cloud usage cells, strict schema/cell checks, D1 migrations and 90-day retention. The page can show configured battery/charger power, battery and solar capacity, generation and recorded grid/battery/vehicle energy. Missing, withheld and stale values stay distinct from zero. Configured limits do not prove physical control. Household hourly collection remains off pending privacy review and policy activation; display that as a waiting state, not a fake zero or collector failure. The updated terms include statistics for every Cloud user; local FTW has a separate Yes/No question with Yes suggested and no upload before an answer.
Private diagnostics require Cloudflare Access plus Worker-side JWT checks. Public dimension labels each need the group minimum. Preserve the legacy signed ingest for compatibility; relay is retired and does not measure current Cloud access.
Validation after merging current main: type checks, 1,133 app tests (two skipped), 24 Worker tests, app build and Worker dry run pass. Resolve the relay type-import conflict by keeping both main's existing Watch exchange and aggregate-export types. Browser review confirms six repos, restored charts, the live estimate and paused energy collection. The deployed Worker is version
04536a74-3012-4ba5-93a6-83691744d2dcfrom commit64188dc; all three D1 migrations are applied.Paired terms changes: srcfl/ftw#1700 (merged), srcfl/ftw-cloud#32, srcfl/ftw-app#87 and srcfl/ftw-web#23 (merged). The original Core/Cloud ingestion foundation is already merged and Cloud is deployed with hourly collection off. This PR does not activate collection or claim a signed DPIA.