From-scratch reimplementation of the Sword of the Stars (2006) engine. Clean-room; engine code only — bring your own game copy. Planning/RE evidence live in sots-re.
a1910becb0
Three phases, one input. A star system carries four per-player masks and three of
them agree on nearly every system of every save the corpus holds, so a model built
on the wrong one looks right until it does not. The gate is the DERIVED
active-presence mask -- fleet-here OR gate-here OR owner, recomputed on every
arrival and departure -- not the sticky one and not the explored one.
S29 SystemObservedStamp the system's own last-observed turn (whole function)
T17 RebuildPlayerViewTree the per-(system, player) observation record: who saw
the system, on what turn, and what encounter was there
T21 UpdateSurveyAndStats the explored sweep: seen this turn implies surveyed
game/sim/visibility is pure and knows nothing about save shapes; app/visibility_phase
wires it to them. The mask is READ FROM THE SAVE and never rebuilt: neither reference
pair moves a mask leaf, so the loaded value is the value these phases would see, and
rebuilding it from an unmodelled movement pass would be a change with no evidence.
Measured, closed and regressed reported separately and never netted:
turn1-state -> turn2-state 209 -> 158 closed 51, regressed 0
turn2-state -> turn3-state 108 -> 87 closed 21, regressed 0
of which this lane closed 46 and 16 (the rest were already closed at main). The 46
are the brief's 32-leaf target in full -- 8 record counts, 8 player ids, 8 turn
stamps, 8 encounter ids -- plus 8 system stamps and 6 explored masks.
Three further pairs the model was never fitted to, all zero regressions:
human-turn2 -> human-turn3 353 -> 311 closed 42 (a different game, 21 systems)
zuul15 -> zuul16 276 -> 264 closed 12
zuul16 -> zuul17 341 -> 329 closed 12
The corpus's one discriminating row is a host test rather than a comment: a system
whose last visiting fleet has gone carries the sticky and explored bits set, the
active bit clear, and a stamp frozen a turn behind. The test asserts the freeze AND
asserts what the wrong gate would have produced, so a future edit that swaps the
mask fails loudly instead of quietly agreeing with five saves.
Labelled hypothesis, with the workload named in the header: the encounter id is
recovered from the encounter fleet at the system, because the field the original
reads is set once at map generation and is not on the wire. It agrees on all six
encounter fleets in the corpus and no save can separate it -- none kills an
encounter while leaving its system visible.
Not written, deliberately: the colony-ownership stamp that moves beside these.
Its gate is demonstrably NOT the active mask (one system in the corpus has a zero
mask and moves it anyway), the formula is not held, so it is reported, not written.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01ARBgSooAfokKUy6wKUKEyZ
|
||
|---|---|---|
| cmake | ||
| docs | ||
| include/generated | ||
| src | ||
| tests | ||
| third_party | ||
| tools | ||
| .gitignore | ||
| CMakeLists.txt | ||
| CMakePresets.json | ||
| CONTRIBUTING.md | ||
| LICENSE | ||
| README.md | ||
sots-engine
A from-scratch, functional reimplementation of the engine behind Sword of the Stars (2006). Not a byte-for-byte decompilation: behavior-equivalent code, built up one verified piece at a time (OpenRCT2-style) until the tree can build the whole application on its own.
Bring your own game. This repository contains engine code only. Game data, assets, saves, and
the original binaries are never included; tests and tools read an owner-supplied copy via
$SOTS_DATA_DIR. See CONTRIBUTING.md for the clean-room rules.
Status
Phase 2. M0 done — the shim frontend (src/shim/, a proxy binkw32.dll the original game
loads) builds, deploys, hooks, and logs from a real game launch. Engine code accrues under
src/mars/ and src/game/; each module is oracle-verified against the owner's game data before merge:
mars/parse— brace-block +.effectreaders (1,531/1,531 files agree with the reference)mars/text— flat key/value tables, id manifests, CSV (64/64 files agree)game/sim— strategic formulas (economy, research, colonies, movement) as pure functions; 356 hand-computed checksmars/vfs—.gob(ZIP) archive reader + loose-file override; entry counts and bytes verified againstunzipmars/stream— the game's self-describing save/serialization format (reader, writer, typed shapes) + gzip; three real saves round-trip byte-identicalmars/rng— MT19937 with save-state load/store; layout confirmed against real saves, draw mappings read off the binary (docs/mars-rng.md)game/data— typed catalogs (weapons, ship sections, turrets, id registries, tech tree, strings) with cross-reference checks; 229k values agree with the referencegame/design— ship-design assembly/fit/tech-gating rules and derived stats; validates all 127 stock designs from real savesapp— the standalone:sots_turnloads a save, runs one strategic turn over the published phase order of all three turn drivers, and writes a save. 14 of the 44 turn-driver phases are modelled; every phase that is not appears in the run log as a named no-op. How far it is from the byte-match, and what stands in the way, is indocs/S-standalone.md
Build: cmake --preset host && cmake --build --preset host && ctest --preset host (Linux);
tools/sync-build.sh cross-builds the shim on the lab box and stages it for deployment.
Layout (grows with the work)
src/shim/— binkw32 proxy + hooks + old-vs-new compare harness (frontend #1)src/mars/,src/game/— the engine and game reimplementation (accruing)src/app/— the standalone turn driver (frontend #2): load a save, run a turn, write a saveinclude/generated/sots_addresses.h— binary facts (RVAs/prototypes), generated from the RE repotests/— host tests; real-data tests skip unless$SOTS_DATA_DIRis settools/— build (MinGW i686 cross) and deploy scripts
Planning, findings, and verification evidence are tracked in the private RE repo (sots-re).