re: the per-phase objectives, from the game's own text

TextIndex in sylpheed-formats::localization already indexes per-phase objective
text under S<NN>_P<n>_Objective_<i>, with hints and lose conditions alongside.
Querying it settles what the wave work spent many iterations circling, and a
disc test now pins it.

Stage 02 phase 1 is "Shoot down all invading enemy fighters while watching out
for attacks on the ACROPOLIS", losing if the ACROPOLIS is sunk. Phase 2 is
"Protect the Admiral's ship the CALIBAN until it has entered the safe zone",
losing if the CALIBAN is sunk. Phase 3 is "Destroy the interstellar cruise
missiles headed for Schlos Base", matching phase 3's roster of exactly nine
ISCMissile and SUBOBJ_013.

That explains the phase advances and shows they are not all the same kind. Phase
1 is a kill-all-marked-fighters gate -- fighters, not the turrets every run
killed almost exclusively. Phase 2 is positional and ends when the CALIBAN
reaches a safe zone, so "what advances a phase" never had a single answer.

It also closes a loop to the first session. The hints say red mission markers
[OB] indicate your targets, so REMAINING OB -- the HUD counter hunted in the
earliest iterations and located at 0xbdb59668 -- is the count of remaining
marked objective targets, which is phase-1 progress itself. The old reframing
that the counter is not a roster was right, and this names what it is. It is
also the correct signal to watch for an advance: not deployed, which only
changes once the next phase deploys, but REMAINING OB reaching zero, which is
the cause.

Method lesson recorded because the reflex will recur: localization.rs has had
objectives, hints and lose_conditions for some time, and several iterations were
spent reconstructing the same information by hand from SUBObjective strings and
the guide script. The reconstruction was not wasted -- it produced the IXUD
UTF-16BE and language\ prefix findings independently -- but the corpus should
have been searched before the disc was, and grep -rl TextIndex crates/ would
have saved the detour.

Minor discrepancy noted, not investigated: ixud.rs documents the string pool as
UTF-16BE and localization.rs as UTF-16LE.

crates/sylpheed-formats/tests/phase_objectives_disc.rs

@@ -0,0 +1,46 @@
+//! Per-phase MAIN objectives, straight from the localisation index.
+//!
+//! The mission-phase work went looking for what a phase asks the player to do
+//! and reconstructed it from `SUBObjective_local_string.tbl` — the *bonus*
+//! objectives — plus the guide script. `TextIndex` already indexes the main
+//! ones under `S<NN>_P<n>_Objective_<i>`, so this pins them as a disc test and
+//! makes the per-phase goals a checkable artifact rather than an inference.
+
+use sylpheed_formats::{localization::TextIndex, PakArchive};
+
+fn pak() -> Option<PakArchive> {
+    let disc = std::env::var("SYLPHEED_DISC").ok()?;
+    PakArchive::open(format!("{disc}/dat/GP_MAIN_GAME_E.pak")).ok()
+}
+
+#[test]
+fn stage02_has_a_main_objective_for_each_phase() {
+    let Some(pak) = pak() else {
+        eprintln!("SYLPHEED_DISC unset - skipping");
+        return;
+    };
+    let text = TextIndex::build(&pak);
+    assert!(text.len() > 0, "text index is empty");
+
+    for stage in ["S01", "S02"] {
+        for phase in 1..=3u32 {
+            let objs = text.objectives(stage, phase);
+            eprintln!("{stage} phase {phase}: {} objective line(s)", objs.len());
+            for o in &objs {
+                eprintln!("    {o}");
+            }
+            for h in text.hints(stage, phase) {
+                eprintln!("    hint: {h}");
+            }
+            for l in text.lose_conditions(stage, phase) {
+                eprintln!("    lose: {l}");
+            }
+        }
+    }
+    // Phase 1 of Stage 02 is the case the wave work could not observe; if the
+    // game states its goal at all, it is here.
+    assert!(
+        !text.objectives("S02", 1).is_empty(),
+        "no main objective text for Stage 02 phase 1"
+    );
+}

docs/re/BACKLOG.md

@@ -560,6 +560,20 @@ search cannot find a *schedule*.
   the pilot (game-playing, uncertain); (2) accept the static answer (structure,
   rosters, routes, objective texts all decoded and cross-confirmed; only the
   *trigger* is inferred); (3) one 40+ min chained run betting against freezes.
+* ✅✅ **(2026-08-24) THE OBJECTIVES, FROM THE GAME'S OWN TEXT** — pinned by
+  `tests/phase_objectives_disc.rs`; `TextIndex::objectives(stage, phase)` already
+  existed. **S02 phase 1: "Shoot down all invading enemy fighters while watching
+  out for attacks on the ACROPOLIS"** (lose: ACROPOLIS sunk). **Phase 2: "Protect
+  the CALIBAN until it has entered the safe zone"** — *positional, not a kill
+  count*, so phase triggers are NOT all the same kind. **Phase 3: "Destroy the
+  interstellar cruise missiles headed for Schlos Base"** (matches the 9
+  `ISCMissile` + `SUBOBJ_013`). ✅ **`[OB]` = objective marker** — hints say *"Red
+  mission markers [OB] indicate your targets"*, so **`REMAINING OB` at
+  `0xbdb59668` is phase-1 progress** and is the right signal to watch, not
+  `deployed`. Closes a loop to the first session. 🔴 **Method lesson: the crate
+  already knew this** — several iterations reconstructed it the hard way;
+  `grep -rl TextIndex crates/` would have saved the detour. ❔ Minor: `ixud.rs`
+  says UTF-16BE, `localization.rs` says LE — one comment is wrong.
   🔴 `SYLPH_HZ=3` refuted as a lever: it gave the LOWEST frame rate (8/s) with the
   most kills, so pilot polling is not the throttle.
 * ~~🚧 BLOCKER: t=210/240 unreachable in one turn~~ — **superseded, see above**;

docs/re/mission-phase-objectives.md

@@ -0,0 +1,71 @@
+# What each phase asks for — from the game's own objective text
+
+Status: ✅ pinned by a disc test (`crates/sylpheed-formats/tests/phase_objectives_disc.rs`).
+
+`TextIndex` in `sylpheed-formats::localization` already indexes per-phase objective
+text under `S<NN>_P<n>_Objective_<i>`, with hints and lose conditions alongside.
+Querying it settles what the wave work spent many iterations circling.
+
+## Stage 02
+
+| phase | objective | lose condition |
+|---|---|---|
+| 1 | **"Shoot down all invading enemy fighters while watching out for attacks on the ACROPOLIS."** | The ACROPOLIS is sunk |
+| 2 | **"Protect the Admiral's ship the CALIBAN until it has entered the safe zone."** | The CALIBAN is sunk |
+| 3 | **"Destroy the interstellar cruise missiles headed for Schlos Base."** | Schlos Base is destroyed |
+
+Stage 01 phase 1 is *"Repel the enemy's surprise attack."*, phase 2 *"Cooperate
+with your allies to repel the enemy units."*
+
+## ✅ This explains the phase advances, and they are not all the same kind
+
+* **Phase 1 is a kill-all-marked-fighters gate.** Not turrets — *fighters*. Every
+  run in [mission-arrival-watch.md](mission-arrival-watch.md) killed turrets
+  almost exclusively, so no run came close to satisfying it.
+* **Phase 2 is positional, not a kill count** — it ends when the CALIBAN reaches
+  a safe zone. That is a completely different trigger type, and it means "what
+  advances a phase" never had one answer.
+* **Phase 3 is destroy-the-missiles**, matching phase 3's roster of exactly nine
+  `UN_e201_ADAN_ISCMissile` and `SUBOBJ_013`.
+
+## ✅ `[OB]` is the objective marker — closing a loop to the first session
+
+The hints name the mechanic outright:
+
+> *"Red mission markers **[OB]** indicate your targets."*
+> *"Take out these enemy units."*
+
+**`REMAINING OB`** — the HUD counter hunted in the earliest iterations of this
+work and located at `0xbdb59668` — is therefore the count of **remaining marked
+objective targets**, i.e. phase-1 progress itself. The reframing recorded back
+then ("the counter is not a roster") was right, and this names what it *is*.
+
+That also makes it the correct signal to watch for a phase advance: not
+`deployed`, which only changes when the next phase deploys, but `REMAINING OB`
+reaching zero, which is the cause.
+
+## 🔴 Method lesson: the crate already knew this
+
+`localization.rs` has had `objectives(stage, phase)`, `hints` and
+`lose_conditions` for some time. Several iterations were spent reconstructing
+the same information the hard way — decoding `SUBObjective_local_string.tbl` by
+hand, reading the guide script, inferring the phase-1 goal from
+*"the attackers with the orange markers"*.
+
+The reconstruction was not wrong, and it produced the IXUD/UTF-16BE and
+`language\` prefix findings independently. But **the corpus should have been
+searched before the disc was**: `grep -rl TextIndex crates/` would have saved
+the detour. Recorded because the same reflex will apply to the next unknown.
+
+Note also a discrepancy worth resolving: `ixud.rs` documents the pool as
+UTF-16**BE** and `localization.rs` as UTF-16**LE**. Both work on their own
+inputs, so one comment is wrong, or the two consumers genuinely differ. Not
+investigated.
+
+## What is still not observed
+
+The phase-1 advance itself. The objective is now known exactly, and
+[mission-objectives-text.md](mission-objectives-text.md) quantifies the cost:
+the marked fighters are ~16 craft and the pilot manages about two per five
+minutes. Nothing here changes that arithmetic — it only means the target list is
+now certain rather than guessed.