Both Dockerfiles (root firebase-mode + docker/ caddy) bumped node:18-alpine to node:22-alpine. better-sqlite3 v12 (from PR #6) requires node 20+; node 18 builds fail with node-gyp/python errors. openssl-legacy-provider flag kept for react-scripts 5 compat. .dockerignore: removed .env* exclusion so .env.local COPY works in firebase image build. mock firestore applyConstraints comment: offset handled by adapter (firebase.js slices), mock never sees it — clarified dead-code reasoning.
3.4 KiB
3.4 KiB
TODO
Backlog of bugs + long-term items. Milestones live in REWORK_PLAN.md.
Open
dm list - keep active particpant in view (scroll)
not sure good way to do this
npm install warnings cleanup pass
lots of updates
TEST GAP: current branch changes need focused coverage
- Storage
where()contract: firebase + server adapters should honor[where('encounterPath','==',x), orderBy('ts','desc'), limit(n)]. - Server SQL query test:
where + orderBy + limitshould return latest logs for one encounter only. - Undo/redo stack order: undo A3 then A2, redo must replay A2 first, then A3.
- Combat controls should not subscribe to logs while mounted; undo/redo should query logs only on click.
- Unified CLI smoke:
node scripts/combat.js verify <fixture.json>returns CLEAN on known-good log. - Unified CLI replay smoke:
node scripts/combat.js replay ... --out tmp/x.jsonwrites JSON array and auto-verifies. - Ctrl-C replay behavior: SIGINT during replay should end encounter, clear active display, write partial log, run verify.
- SQLite schema/index test:
idx_docs_parent_tsandidx_docs_parent_encounter_tsexist for server DB.
confirm warnings treated as error = fail in all tests, build pipeline, linters, everything. again.
BUG: addParticipants (batch add) does not slot by initiative
- shared/turn.js addParticipants appends
[...existing, ...new], no slotIndexForInit. - Violates INIT doc: "Add = insert into slot by initiative."
- Single addParticipant slots correct. Batch (add-all-chars) appends.
- Pre-start batch add = wrong order. Post-start worse.
BUG: nextTurn throws on solo combatant
- nextActiveAfter loop
for step=1; step<nskips when n=1 → {nextId:null}. - nextTurn throws "Could not determine next participant."
- Solo active combat cannot pass turn.
BUG: reorderParticipants cross-pointer drag = silent no-op
- Cross-pointer drag returns encounter unchanged, no log, no toast.
- DM drags across current turn → nothing happens, no feedback.
BUG: addParticipant undo missing currentTurnParticipantId when started
- undo saves participants + conditional turnOrderIds, no currentTurnParticipantId.
- Pointer can misalign on undo if added near pointer region.
BUG: computeTurnOrderAfterRemoval isActive uses find() not boolean
isActive = id => updatedParticipants.find(p => p.id === id && p.isActive)- Returns participant obj (truthy) not boolean. Works by accident, fragile.
BUG: select campaign during active combat = screen flicker, no action
- In active encounter, click different campaign → flicker, no nav, no end.
- Either block (toast: end encounter first) or auto-end current + switch.
- Decide UX before fix.
FEAT: generic/non-5e rules mode
- Campaign/encounter ruleset toggle:
5evsgeneric. - Generic mode turns off death saves/state machine.
- Generic mode allows negative HP.
- 5e mode rejects negative damage/heal and clamps HP at 0.
FEAT: clarify "Is NPC" in add-participant
- Ambiguous label. May expand work based on what NPC means here (ally? monster? display-only? skip in turn order?). Clarify intent before UX changes.