trx-rs/WIP.md

WIP: Per-Rig Bookmarks, Scheduler, Decoders, and Map Rig Selector

Goal

Make bookmarks and scheduler per-rig, filter decoders tab by current rig, and use remote names in the map rig selector.

---

1. Per-Rig Bookmark Storage

Current: Single ~/.config/trx-rs/bookmarks.db shared by all rigs.

Target: One file per rig: ~/.config/trx-rs/bookmark.{remote_name}.db (e.g. bookmark.ft817.db, bookmark.lidzbark-vhf.db).

Changes

• bookmarks.rs — BookmarkStore::default_path() → BookmarkStore::default_path(remote: &str). Replace the single static path with a method that embeds the remote name.
• server.rs — Instead of opening one BookmarkStore, maintain a HashMap<String, Arc<BookmarkStore>> keyed by rig id (lazily opened). Provide a helper bookmark_store_for(remote: &str) -> Arc<BookmarkStore>.
• api.rs bookmark endpoints — All bookmark REST endpoints gain a required ?remote= query parameter. Route to the corresponding per-rig store. Return 400 if remote is missing or unknown.
• bookmarks.js — Pass the active rig id (lastActiveRigId) as ?remote= on every fetch/POST/PUT/DELETE. Reload bookmarks on rig switch.
• Scheduler integration — spawn_scheduler_task and BackgroundDecodeManager already receive bookmark_store; update them to resolve the per-rig store from the rig id they're operating on.

Migration

On startup, if the legacy bookmarks.db exists and no per-rig files exist yet, copy it to every configured rig as a one-time migration, then rename the old file to bookmarks.db.migrated.

---

2. Per-Rig Scheduler Storage

Current: Single ~/.config/trx-rs/scheduler.db with keys sch:{remote}.

Target: One file per rig: ~/.config/trx-rs/scheduler.{remote_name}.db.

Changes

• scheduler.rs — SchedulerStore::default_path() → SchedulerStore::default_path(remote: &str). Same pattern as bookmarks. Remove the sch: key prefix since each file is already scoped to one rig. Use a single key like config instead.
• server.rs — Maintain HashMap<String, Arc<SchedulerStore>> keyed by rig id (same lazy-open pattern).
• api.rs scheduler endpoints — Already take {remote} path param; route to the per-rig store instead of the single shared store.
• scheduler.js — No URL changes needed (already uses {remote} in path). Verify bookmark loading uses per-rig bookmarks.

Migration

Same pattern: if legacy scheduler.db exists and no per-rig files exist, extract each sch:{remote} entry into its own scheduler.{remote}.db, then rename old file.

---

3. Decoders Tab — Filter by Current Rig

Current: Decode history and live SSE stream are global (all rigs mixed).

Target: Decoders tab shows only decodes from the currently selected rig.

Changes

• decode.rs / audio.rs — Check if decoded messages already carry a rig_id field. If not, tag each decode event with the originating rig id when it enters the history ring buffer.
• api.rs /decode/history — Accept optional ?remote= query param. When present, filter history to only that rig's decodes.
• api.rs /decode SSE — Accept optional ?remote= query param. When present, filter live events to that rig.
• app.js — When connecting to /decode SSE and fetching /decode/history, append ?remote={lastActiveRigId}. Reconnect/refetch on rig switch.
• decode-history-worker.js — No changes needed if server-side filtering is applied (worker just decompresses what it gets).

Decode message tagging

If Ft8Message, AprsPacket, etc. in trx-core/src/decode.rs don't already have a rig_id field, add an optional rig_id: Option<String> to each type. The server fills this when the decoder produces output. The history collector preserves it. Filtering is done at query time.

---

4. Map Rig Selector — Use Remote Names

Current: updateMapRigFilter() in app.js populates #map-rig-filter options using lastRigDisplayNames[id] || id. The display_name comes from RemoteRigEntry.display_name which is the manufacturer/model string.

Target: Use the remote name (the rig id from config, e.g. ft817, lidzbark-vhf) as the option value (already done), and show the display_name if available (already done). The user wants to ensure remote names are used consistently — verify and fix any places that fall back to manufacturer/model instead of the configured remote name.

Changes

• app.js updateMapRigFilter() — Already correct: uses id (remote name) as value and lastRigDisplayNames[id] || id as label. Verify this uses the remote name, not some internal index.
• api.rs /rigs — Verify remote field is the config remote name (rig id), not a generated identifier. Should already be correct.
• If display_name is not set, fall back to remote name (not "manufacturer model").

---

Implementation Order

1. Per-rig bookmark storage (Rust backend + JS frontend + migration)
2. Per-rig scheduler storage (Rust backend + migration)
3. Decoders tab rig filtering (decode message tagging + API filtering + JS)
4. Map rig selector remote name cleanup (JS)

---

Files to Modify

Rust

• src/trx-client/trx-frontend/trx-frontend-http/src/bookmarks.rs
• src/trx-client/trx-frontend/trx-frontend-http/src/scheduler.rs
• src/trx-client/trx-frontend/trx-frontend-http/src/server.rs
• src/trx-client/trx-frontend/trx-frontend-http/src/api.rs
• src/trx-client/trx-frontend/trx-frontend-http/src/audio.rs
• src/trx-core/src/decode.rs (if rig_id tagging needed)

JavaScript

• src/trx-client/trx-frontend/trx-frontend-http/assets/web/app.js
• src/trx-client/trx-frontend/trx-frontend-http/assets/web/plugins/bookmarks.js
• src/trx-client/trx-frontend/trx-frontend-http/assets/web/plugins/scheduler.js