Roadmap
The strategy is a vertical slice: build the hardest, most-differentiating piece (a tactical battle) first, make it fun, then grow outward into a campaign.
Milestones
| Milestone | Scope | Status |
|---|---|---|
| M1 | One playable, self-contained tactical battle. | ✅ Scaffolded & playable |
| M2 | A dynastic campaign map — clickable provinces, characters/realms, turn-based diplomacy & war. Battles auto-resolved at first. | 🚧 In progress — the Gallic War map, data-driven campaigns, and diplomacy (with AI agency) have landed; characters/dynasty and the saga layer remain |
| M3 | Integration — an army battle on the campaign map launches into the M1 battle scene and returns a result (winner, casualties). Where the two genres meet. | 🚧 In progress — the battle hand-off and result-applied-to-map round trip have landed; AI-vs-AI/AI-vs-player battle launches and richer army composition remain |
The battle scene was kept self-contained specifically to make the M3 hand-off clean.
Prioritized backlog
The foundation, the collision pillar (per-type footprint, pathfinding, formation cohesion, hard blocking), and the shared collision-exemption primitive (friendly pass-through, line relief, unit merging) all shipped — see PLAN.md’s design-pillars section for how each landed. With that early backlog cleared, the issue tracker itself is the live source of truth for what’s prioritized now: browse open issues on Lacaedemon/sparta with P0–P3 labels, or the dominant current threads below.
Three threads dominate the current backlog, feeding the M2/M3 milestones above and the battle presentation:
- Campaign layer — phase 1 through phase 7 build out map/time, army composition, battle-resolution modes, real-geography ingestion, logistics, sagas, and dynasty in sequence. Phase 1’s foundation has begun landing: the campaign map now carries a pausable clock with speed controls and a day counter (see How to Play), and each campaign declares how much ground a map unit covers, so a completed march reports how far the army walked in kilometres. The projection that phase 4’s real-geography ingestion will author against ships alongside it. Continuous army movement on that substrate, and retiring turn order in favour of the clock, are the rest of the phase.
- Battle AI — subcommanders, the general’s doctrine/reserves layer, player delegation to AI subcommanders, and fog-of-war-aware perception (phases 2–5) have all shipped; see Tactics for what’s live and Tactics for how fog now shapes both sides’ decisions, closing out the chain-of-command build-out. Phase 5 — the fog-of-war-aware perception swap — spent over two weeks closed in error and was reopened on 2026-09-19, Pacific (see
docs/battle-ai-design.md’s “Implementation status” section for why). Commander-scoped narrowing (a subcommander’s own view, rather than a team-wide one) and up-the-chain report propagation are still open, tracked as #1625; AI reasoning over a lost contact’s last-known position (the same memory the player’s own ghost markers already show) is tracked separately as #1626. - 3D battlefield — a planned conversion of the battle view from 2D top-down to a Total War-style 3D presentation with a rotatable camera (#69); the deterministic simulation stays planar and only the render/camera/input shell changes. The architecture and phased plan live in
docs/3d-conversion-design.mdin the repository.
Independent of those threads, P2/P3 also carries near-term fixes and design work (formation/spawn edge cases, the soldier weapon/shield type model, a maneuver-queue refactor) — the tracker’s labels and open-issue list are more current than any snapshot repeated here.
Source: PLAN.md in the repository. The issue tracker is the live source of truth for status.