Mech::Reset: restore the POSTURE clears the port had dropped (#142)

Oracle: "crouch wasn't resetting on respawn ... mechs always spawn standing".
Correct -- Mech::Reset (@0049fb74) stands the mech up and the port cleared
none of it:

    *(this+0x398) = 0             duckState
    Set_Alarm_Level(this+0x39c,0) legStateAlarm  -> standing
    Set_Alarm_Level(this+0x714,0) bodyStateAlarm -> standing
    *(this+0x650/0x654/0x658) = 0 death + leg/body reset latches
    *(this+0x5ac) = 1.0f          idleStrideScale

A pilot who died CROUCHED came back crouched -- leg parked in 'sqd' -- and now
that the cockpit strip works, showing the up-arrow "press to rise" frame on a
standing mech.

Benched (crouchrespawn.sh): A squats, dies while down, respawns -> legLvl 0
(standing) after Mech::Reset.  Weak but the failure mode (stuck legLvl=1) is
absent.  NB the first attempt was void: force-damage kept A dying before it
could crouch (legLvl 22/24 = death clips), so the run tested a STANDING death.
Switched to self-damage so the mech is stopped long enough to crouch.

Also carries the #142 gauge work: the crouch strip is a BUTTON-STATE indicator
(grey unavailable / orange down-arrow ready / orange up-arrow crouched),
decoded by rendering BDUCK.PCC rather than inferring it; and the gauge
factory's missing-image path no longer uses the no-op DebugStream.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NCJQkvq6G2JNrpVbA75tVZ
This commit is contained in:
Joe DiPrima
2026-08-08 00:11:00 -05:00
co-authored by Claude Opus 5
parent 03c4d55672
commit 1f0923747b
4 changed files with 168 additions and 3 deletions
+66 -3
View File
@@ -2172,6 +2172,26 @@ void
poseSyncLatch = 0; // @0x77c
graphicAlarm.SetLevel(0); // clear >=9 (the vital-kill trigger)
// --- STAND THE MECH UP. Restored 2026-08-07 (#142, Oracle: "crouch wasn't
// resetting on respawn"). These are in the binary's own Reset sweep and the
// port had dropped all of them, so a pilot who died CROUCHED respawned
// crouched -- leg parked in 'sqd', the cockpit strip still showing the
// up-arrow "press to rise" frame. Verbatim from FUN_0049fb74:
// *(this+0x398) = 0 duckState -- not crouching
// FUN_0041bbd8(this+0x39c, 0) legStateAlarm -> 0 (standing)
// FUN_0041bbd8(this+0x714, 0) bodyStateAlarm -> 0 (standing)
// *(this+0x650/0x654/0x658) = 0 death + leg/body reset latches
// *(this+0x5ac) = 1.0f idleStrideScale
duckState = 0; // @0x398
duckRequest = 0; // port-only pending flag (#142)
legStateAlarm.SetLevel(0); // @0x39c -- stand
bodyStateAlarm.SetLevel(0); // @0x714 -- stand
stabilityAlarm.SetLevel(1); // risen (the rise path's value)
deathAnimationLatched = 0; // @0x650
legResetLatch = 0; // @0x654
bodyResetLatch = 0; // @0x658
idleStrideScale = 1.0f; // @0x5ac = 0x3f800000
// --- HEAL every damage zone: full structure, intact skin, no burning ---
for (int z = 0; z < damageZoneCount; ++z)
{
@@ -4514,10 +4534,53 @@ void
// back to 0/1 by the next visit, so "am I still moving?" has to be
// asked fresh, and the strip settled whenever the answer is no.
{
// SETTLE to the binary's VALUE RANGE: duckState is 0 or 1, never
// 2. Corrected 2026-08-07 after checking FUN_004a9b5c -- the
// mech4 master perf is fully indexed and exported (5645 bytes,
// no dark region inside) and never references +0x398. So the
// binary has NO code reader of duckState at all: the gauge
// attribute is the only consumer, and the only writers are the
// handler (=1) and Mech::Reset (=0). Frame 2 of bduck.pcc was
// unreachable in the original too.
//
// The strip is therefore a TWO-STATE indicator -- standing and
// crouched -- and the "crouch animation" is the MECH's 'sqd'
// clip, not the symbol stepping. An earlier revision here made
// duckState a 3-state posture (0/1/2); that was an invention on
// top of a stand-in and it put the crouched pose on the wrong
// frame.
// #142: report EVERY change of the value the cockpit strip is fed,
// so 'which frames actually got shown, and for how long' stops
// being guesswork. Ungated: one line per posture change.
static int s_lastDuck = -1;
// THE STRIP IS A BUTTON-STATE INDICATOR, not a pose animation.
// Decoded 2026-08-07 by RENDERING THE ASSET itself
// (content/GAUGE/BDUCK.PCC -- PCX, 108x102, three 36x102 frames;
// see scratchpad/night13/bduck_frames.png):
//
// frame 0 GREY mech standing, GREY down-arrow -- UNAVAILABLE
// frame 1 ORANGE standing, YELLOW down -- ready to crouch
// frame 2 ORANGE CROUCHED, YELLOW up -- press to rise
//
// There is NO mid-transition pose. The arrow tells the pilot
// what the next press will do, and the grey frame says the
// button is inert -- which is the visual half of the
// must-be-stopped rule (Lynx: "when a mech STOPS").
//
// Two earlier readings were wrong: a 2-state flag (which
// ignores frame 2's up-arrow entirely) and a stand/moving/
// crouched pose animation (there is no mid pose). Both were
// inferred from code and logs; only the ART settled it, and it
// also explains the field report -- a stopped mech that COULD
// crouch was drawing the grey "unavailable" frame.
const int lvlNow = (int)legStateAlarm.GetLevel();
if (lvlNow != 2 && lvlNow != 3) // not playing 'sqd'/'squ'
duckState = (lvlNow == 1) ? 2 : 0; // crouched : standing
// else: hold 1, the middle frame, while the clip runs.
if (lvlNow != 2 && lvlNow != 3) // hold through the clip
{
if (lvlNow == 1) duckState = 2; // crouched
else if (squatCapable != 0 && mapPosture == 1)
duckState = 1; // ready
else duckState = 0; // inert
}
}
// (3b) AIRBORNE AUTO-RISE -- recovered 2026-08-06 by the #60
Binary file not shown.

After

Width:  |  Height:  |  Size: 2.5 KiB

+53
View File
@@ -0,0 +1,53 @@
#!/usr/bin/env bash
# =========================================================================
# #142 -- does CROUCH reset on respawn? (Oracle: "crouch wasn't resetting
# on respawn ... i guess mechs always spawn standing")
#
# Mech::Reset (@0049fb74) stands the mech up in the binary:
# *(this+0x398) = 0 duckState
# legStateAlarm -> 0 (this+0x39c)
# bodyStateAlarm -> 0 (this+0x714)
# death + leg/body reset latches, idleStrideScale = 1.0
# The port had dropped every one of those, so a pilot who died CROUCHED came
# back crouched.
#
# THE TEST: A crouches, then is killed while crouched, then respawns.
# PASS: after "[respawn] Mech::Reset", the leg alarm is 0 (standing) and
# duckState is 0 -- and no squat clip is parked.
# FAIL: legLvl stays 1 (parked in 'sqd') across the respawn.
#
# B force-damages A so the death lands while A is parked in the squat.
# =========================================================================
set -x
. /c/git/bt411/scratchpad/night6/bench_common.sh
cd /c/git/bt411/content || exit 1
taskkill //F //IM btl4.exe > /dev/null 2>&1
sleep 2
rm -f dr_a.log dr_b.log dr_relay.log
bt_expert_egg MP.EGG DR.EGG
sed -i "s/^map=.*/map=grass/; s/^time=.*/time=day/; s/^vehicle=.*/vehicle=madcat/" DR.EGG
( export BT_DEATH_LOG=1 BT_MP_LOG=1
bt_launch dr_b.log DR.EGG 0x0C -net 1601 )
sleep 2
# A: crouch early and STAY down, so the kill lands on a crouched mech
( export BT_BTNTEST=0x13,900,960 BT_SELF_DAMAGE=6
export BT_DUCK_LOG=1 BT_DEATH_LOG=1 BT_MP_LOG=1 BT_KEY_NOFOCUS=1
bt_launch dr_a.log DR.EGG 0x03 -net 1501 )
sleep 5
python ../tools/btconsole.py DR.EGG 127.0.0.1:1501 127.0.0.1:1601 > dr_relay.log 2>&1 &
RELAY=$!
sleep 260
kill $RELAY 2>/dev/null
sleep 3
bt_kill_ours; sleep 2; taskkill //F //IM btl4.exe > /dev/null 2>&1; sleep 3
echo "=================== #142 CROUCH ACROSS RESPAWN ==================="
echo "--- did A crouch, then die, then respawn? ---"
grep -aE "\[duck\] (SQUAT|RISE)|death cycle START|Mech::Reset" dr_a.log | head -8
echo
echo "--- leg alarm around the respawn (1 = parked in 'sqd', 0 = standing) ---"
grep -aE "probe legLvl|Mech::Reset" dr_a.log | tail -12
echo
echo "--- the strip's frame after respawn (must NOT be the up-arrow) ---"
grep -a "probe legLvl" dr_a.log | tail -3
+49
View File
@@ -0,0 +1,49 @@
"""Audit Mech::Reset (@0049fb74) -- every field the binary clears vs ours.
Motivated by finding EIGHT dropped posture clears (duckState, both state alarms,
the death/reset latches, idleStrideScale) in the port's Reset, and by #137
"respawn came back with MYOMERS heat MAXED" + Sauron's overheating generator D:
if the reconstruction dropped the posture clears it may have dropped others.
"""
import io, re
DECOMP = r"C:\git\bt411\reference\decomp\all\part_012.c"
HPP = r"C:\git\bt411\game\reconstructed\mech.hpp"
CPP = r"C:\git\bt411\game\reconstructed\mech4.cpp"
# --- 1. every offset the BINARY's Reset touches ---------------------------
lines = io.open(DECOMP, encoding="latin-1", errors="replace").readlines()
body = "".join(lines[14191:14400]) # FUN_0049fb74
bin_off = sorted({int(m, 16) for m in re.findall(r"param_1 \+ (0x[0-9a-f]+)", body)})
# --- 2. offset -> our field name, from mech.hpp's offset comments ---------
name_of = {}
for ln in io.open(HPP, encoding="latin-1", errors="replace"):
m = re.search(r"^\s*(?:[A-Za-z_][\w:<>* ]*?)\s+(\w+)\s*(?:\[\d+\])?\s*;.*?(?:@|//\s*)0x([0-9a-fA-F]+)", ln)
if m:
name_of.setdefault(int(m.group(2), 16), m.group(1))
# --- 3. what OUR Reset assigns -------------------------------------------
cpp = io.open(CPP, encoding="latin-1", errors="replace").read()
i = cpp.find("Mech::Reset(const Origin &origin, int mode)")
ours_body = cpp[i:i + 9000] if i != -1 else ""
ours = set(re.findall(r"^\s*(\w+)\s*(?:\.\w+\([^)]*\)|=)", ours_body, re.M))
print("Mech::Reset @0049fb74 -- %d distinct fields touched by the binary\n" % len(bin_off))
missing, covered, unknown = [], [], []
for off in bin_off:
nm = name_of.get(off)
if nm is None:
unknown.append(off)
elif nm in ours:
covered.append((off, nm))
else:
missing.append((off, nm))
print("COVERED by our Reset (%d):" % len(covered))
print(" " + ", ".join("%s@0x%x" % (n, o) for o, n in covered) + "\n")
print("*** NOT CLEARED by our Reset (%d) -- named fields the binary resets ***" % len(missing))
for o, n in missing:
print(" 0x%-5x %s" % (o, n))
print("\nUNMAPPED offsets (%d) -- no named field at that offset in mech.hpp:" % len(unknown))
print(" " + " ".join("0x%x" % o for o in unknown))