feat(ui): render LiveValue.text enum labels in gauge + CSV

[ohgeeceee]

Wires the frontend half of the u8_enum feature shipped to the backend in PR #60 (commits 643bb24, 995a209). With this change, opening Live Data with a B58 profile renders the new gear and engine_state parameters as text gauges ("3rd", "Running") instead of as a 0.0 needle pinned to min.

What ships

• src/js/gauges.js — Gauge gains a text-mode rendering path:
  • set(value, label?): when label is non-null, the gauge enters text mode; otherwise it behaves exactly as before.
  • tick(): skips the easing loop in text mode (labels are discrete, not numeric).
  • draw(): renders the dial face (kept for visual consistency with adjacent numeric gauges), then either the text label centred with the unit underneath, or the existing dial / ticks / needle / numeric readout. Long labels auto-shrink the font from 22 → 12 px so words like "Cranking" or "Overrun" fit without clipping.
  • textOverride field on the instance; defaults to null so numeric gauges are unchanged.
• src/js/main.js:
  • pollOnce (line 720): ensureGauge(v).set(v.value, v.text).
  • logTick (line 1347): stores point.text alongside y for each logged sample.
  • buildLogCsv (line 1584): emits the label as a quoted CSV cell (JSON.stringify(p.text)) when present, falling back to the existing two-decimal numeric format otherwise. Result: a B58 gear log exports 0.00,"P/N",0.00,"1",... instead of 0.00,0,0.00,1,....
• CHANGELOG.md: [Unreleased] → ### Added entry referencing PR feat(data): u8_enum decoder + per-parameter enum-map pipeline #60.

Protected-path flags

None touched. transport/, protocol/, commands.rs, and bmw_diag/core/** are all unchanged. This is a frontend-only PR.

Scope deliberately deferred

• Chart rendering for enum params — categoricals over time (e.g. a gear line graph that steps from 0 → 1 → 2 → 3 → 4) are a meaningful UX problem worth their own design pass. The chart still renders an enum param as its numeric value (always 0.0 today) when a user opts in via the existing checkbox. No regression; no new behaviour here.
• Parameter Explorer enum editor — letting users author enum maps in-app is a UX-heavy follow-up. Use TOML for now.
• Reformatting drift in the wider repo. Pre-existing.

Verification

• cargo test --lib — 34/34 passing (no Rust touched; verified on clean origin/main first).
• node --check src/js/main.js src/js/gauges.js — clean syntax.
• Manual visual check: the system reminder flagged this is creative UI work; I have NOT launched Tauri to confirm the gauge canvas renders the label text correctly across font-size and label-length combinations. The auto-shrink loop has been tested by reading the math: at 22 px with the default r=74, the gauge accepts labels up to ~118 px wide before shrinking. Real-world BMW labels ("P/N", "1", "Cranking", "Running", "Idle", "Overrun", "Shutdown") are all under 90 px in 22 px Segoe UI, so no shrink should be needed in practice.

Refs

• Backend: PR feat(data): u8_enum decoder + per-parameter enum-map pipeline #60 (643bb24 + 995a209), LiveValue.text field, decode_enum_string.
• docs/DECODE_FUNCTIONS.md §8.
• UNIQUE_FEATURES.md §1 (the eventual Community Oracle stack).
• Plan: .hermes/plans/u8_enum_frontend_plan.md (gitignored working doc).

Risk + rollback

Single revert, no migrations, no schema, no async surfaces touched. Rollback restores the gauge to numeric-only and the CSV to numeric-only exports.

[ecc-tools]

ECC bundle files are already tracked in this repository. Skipping generation of another bundle PR.

[chatgpt-codex-connector]

💡 Codex Review

beemuu/ops/backup-beemuu-db.sh

Line 36 in a708c54

DB_PATH=""

Avoid overwriting the resolved DB path

This new resolver is dead code in the normal production case: after it computes BEEMUU_DATA_DIR or /var/www/beemuu/data/beemuu.db, the assignment below resets DB_PATH back to ${BEEMUU_DB_PATH:-/var/www/beemuu/backend/data/beemuu.db}. On hosts using the documented new /var/www/beemuu/data layout without BEEMUU_DB_PATH, the backup still looks at the old backend path and exits; please remove the later reassignment or fold the resolver into it.

---

beemuu/src-tauri/src/data/live.rs

Line 71 in a708c54

pub enum_map: std::collections::HashMap<u8, String>,

Round-trip enum maps when exporting profiles

With this new per-parameter enum_map, exporting a community profile that contains u8_enum params through export_profile/profile_to_toml loses the map: the TOML still emits decode = "u8_enum" but no enum = { ... }, so importing that export creates empty maps and read_live_data never emits those enum values. Please serialize non-empty enum maps in profile_to_toml so B58/N55 exports remain usable.

---

beemuu/src-tauri/src/commands.rs

Line 259 in a708c54

continue;

Keep unknown enum bytes visible

With the new best-effort enum profiles, any raw byte that is missing from enum_map takes this continue path without pushing a LiveValue. In those unverified/new ECU states the gauge and logging row disappear instead of showing an unknown/raw value, which prevents users from seeing or capturing the byte needed to extend the map; please emit a fallback text/raw value when decode_enum_string returns None.

---

beemuu/src/js/main.js

Line 181 in a708c54

if (advOpen && kind === "kdcan") await refreshPorts();

Refresh K+DCAN ports even when options are collapsed

When a user switches from Simulator to K+DCAN while the new connection-options panel is collapsed, advOpen is false so this no longer calls refreshPorts(). The hidden conn-port select then stays empty, and clicking Connect sends an empty port to the K+DCAN transport instead of the available serial device; the old handler refreshed ports on every K+DCAN selection. Please refresh the port list regardless of the collapsed state, or before connecting.

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".

[ecc-tools]

ECC bundle files are already tracked in this repository. Skipping generation of another bundle PR.