sots-re/findings
alex 49fa5501ba lane T2: the treaty-turn stamp, and Player.Status's real writer
findings/subsystems/treaty-turn-stamp.md. Briefed on the post-combat tail, ranked its
phases by leaves-closed-per-effort against the standalone's 158-leaf residual, and found
that every tail phase which moves a leaf in this corpus is blocked on another lane
(ship construction, the budget) or on the tail's vtable blind spot. Section 5 lists that
ranking with the numbers. The two items below are what a lane holding no VM could close.

1. StrategyServer_StampTreatyTurns 0x007898c0, read as instructions. The diplomacy
   ledger's per-turn stamp over every ordered pair of players holding a treaty, creating
   the entry on demand. Sole caller is ApplyTurnCommands, so it is a HOST step and not a
   tail phase. Three corrections to strategic-turn-internals.md 5.2, which is flagged in
   place: the relation codes are 3=ally / 2=NAP / 1=cease-fire and not the reverse (which
   combat-done-tail.md 2A already implied independently); the stamped value is Frame, not
   ModCount; and the bit is the PlyrIdx field, not the player's vector position -- the
   opposite convention from the shared-vision mask.

   Modelled in sots-engine as host phase H02. Predicted 26 leaves on the reference pair
   and 14 on pair 2 before building; measured exactly 26 and 14, 0 regressed, and +12 on
   each of three further pairs from a different game at turns 2, 15 and 16 that the model
   was never fitted to. 76 closed / 0 regressed over five pairs, 0 RNG words. The rule
   also reproduces the ledger of ten of the eleven corpus saves entry for entry.

2. Player.Status. backlog.md item 6 has it blocked on "the writer between tail 31 and the
   autosave -- watchpoint". There is no such writer. A whole-image scan for immediate
   stores to ServerPlayer+0x164 finds exactly three: ProcessTurn's encounter loop writes
   1, ResumePlaying writes 0 on load (the 4 -> 0 the determinism note recorded), and
   StrategyServer_MarkPlayerTurnEnded 0x00821a40 writes 4 from the three End Turn
   SUBMISSION paths, which run before the turn is processed. Item 6 needs one predicate,
   not a watchpoint -- and an entry probe on that function names the set directly.

   NOT written. The only two saves in the corpus with a non-zero Status agree with
   Species != 4 on all eight players, and that is eight observations on a 1-bit predicate
   that nine of the eleven saves do not exercise at all. Evaluated and reported.

Seven addresses in ghidra/addresses.d/lane-t2.json, validated to a scratch path against
the merged set (1,095 entries, no duplicate). verify/results/standalone/{report,status}
deliberately untouched: lane C3 published into them minutes before this run.
2026-09-08 14:52:44 -04:00
..
control-flow lane H: five live probes, the oracle restored, and a harness perturbation bisected to one hook 2026-09-08 13:53:36 -04:00
objects lane D2: the ship-design catalogue -- how designs persist, hull size, and the 0x400 flag 2026-09-08 12:45:36 -04:00
subsystems lane T2: the treaty-turn stamp, and Player.Status's real writer 2026-09-08 14:52:44 -04:00
.keep.md init commit 2026-09-07 12:31:56 -04:00
00-strategy.md strategy: phase 2 decisions (two repos, c++ shim, accrual order) 2026-09-07 17:10:21 -04:00
01-fingerprint.md map: RTTI class inventory (1924 types, engine=Mars); reseed board with real targets 2026-09-07 14:48:43 -04:00
_template.md scaffold RE campaign: board, backlog, findings template, verify tree 2026-09-07 14:40:44 -04:00