Game

The Legend of Zelda: The Wind Waker

Nintendo's 2002 GameCube Zelda, sailed across an ocean by a wind the player turns, read through the zeldaret decompilation.

stubAlso: The Wind Waker, Wind Waker, TWW, Zelda no Densetsu: Kaze no TakutoUpdated 2026-10-22
On this page
  1. Related
  2. Mechanics
  3. References
  4. Sources

The atlas reads The Wind Waker through the Zelda Reverse Engineering Team’s decompilation, C++ that rebuilds each GameCube release; the claims here cite the code the USA release compiles. Its boat actor doesn’t yet compile to the game’s bytes as a whole, so the boat claims read the decompilers’ reconstruction. Names and comments quoted here are theirs. Nothing has been measured in the game yet.

The boat’s speed comes from the wind. The code compares the wind’s direction with the boat’s facing: with the wind along the heading the sail gives its full top speed of 55, across it three quarters, at 135° under a third, and past about 140° nothing1. With the sail down, the stick moves it at up to 10 instead1. The tiller turns it fastest at rest and under a third as fast at full sail2. With the sail up and enough speed, R hops the boat3.

The wind has a strength from the room and a direction from the player. Each room’s wind level gives 0.3, 0.6, or 0.9, and the boat reaches its top speed only at 0.5 or more4. The Wind’s Requiem points the wind one of eight ways, 45° apart, and the save keeps the choice5.

The clock counts a day as 360 units and adds 0.02 an update in rooms where time passes; in other rooms it runs only until 11:00 by day or 23:00 by night6.

See also

Mechanics

MechanicKindHow
SailingWind anglethe angle between the wind and the boat's facing scales a top speed of 55: full with the wind along the heading, three quarters across it, none past 140°
WindPlayer-setthe Wind's Requiem picks one of eight directions, 45° apart, kept in the save; each room's wind level sets the strength
Time of dayGame clocka day is 360 units, 15 an hour, advancing 0.02 an update where time passes; elsewhere it creeps only to 11:00 or 23:00
Every claim about this game, by concept (6)
SailingIn the decompilation's C++ for the boat (d_a_ship, which doesn't yet match the game's bytes as a whole; the USA GZLE01 build), procSteerMove, while the boat isn't airborne (daSFLG_FLY_e off), takes the gap between the wind vector's yaw (cM_atan2s of its x and z) and the boat's facing shape_angle.y, as a 16-bit angle from 0 to 0x8000, and turns it into a wind factor: 1 − gap/0x10000 below 0x4000 (90°), 0.75 − (gap − 0x4000) × 5.493164e-05 up to 0x6000 (135°), then 0.3 − (gap − 0x6000) × 0.00032967035, floored at 0. So the factor is 1 with the wind along the boat's facing, 0.875 at 45°, 0.75 at 90°, 0.3 at 135°, and 0 from 140° to 180° (computed). The target speed is the wind power divided by 0.5, capped at 1, times the factor times 55. The branches then run in this order, one per update. First, on the body animation's frame 7, if speedF is below twice the target (at most 80) it is raised to that and mTargetSpeed is set to the target (10000 if the boat is being moved by force), and nothing else changes speedF that update. Otherwise, while mTargetSpeed is below speedF, firstDecrementShipSpeed lowers speedF toward mTargetSpeed (a tenth of the gap, at most 5, at least 1) and sets mTargetSpeed to 10000 once the gap is under 3; this comes before the sail check, so it applies with the sail flag off too. Otherwise, with the sail-on flag off or a target of 0, decrementShipSpeed eases speedF toward 10 × the stick's value × the cosine of its angle (a twentieth of the gap, at most 0.1, at least 0.015). Otherwise decrementShipSpeed eases it down to a lower target, and cLib_addCalc up to a higher one (a tenth of the gap, at most 0.5, at least 0.1).Code
In the decompilation's C++ for the boat (d_a_ship, not yet matching as a whole; the USA build), procSteerMove moves the unnamed angle m0366, which the atlas reads as the tiller, toward mStickMVal × 0x2000 × −sin(mStickMAng) with cLib_addCalcAngleS(…, 4, l_HIO.tiller_speed, 0x100), tiller_speed being 700, and, unless the boat is airborne, passes it to setMoveAngle. Unless the boat is being moved by force (checkForceMove), setMoveAngle sets a rate of 4 − |speedF| / 55 × 3, held between 0.1 and 3.6, blends it toward 3.6 by the unnamed counter m0388 over 15 (3.6 once m0388 reaches 15), and turns the facing by −rate × (m0366 >> 6). With m0388 at 0 and a tiller of 0x2000, that is about 461 units (2.5°) an update at rest and 128 units (0.7°) at speed 55 (computed): the boat turns slower the faster it sails.Code
In the decompilation's C++ for the boat (d_a_ship, not yet matching as a whole; the USA build), procSteerMove allows a hop when no event runs, Link isn't barred from control (checkNoControll), the body animation is FN_MAST_ON2 at frame 7 or later, the boat isn't airborne, its forward speed mFwdVel is over 16.5, it isn't being moved by force, and the sailing target speed is at least 11. Then the R trigger's lock (mDoCPd_R_LOCK_TRIGGER) sets the airborne and jump flags and an upward speed equal to the target speed, held between 15 and 40, and sets mJumpRate to (speed.y − 11) × 0.014545455.Code
Time of dayIn the USA build, the time of day is a float from 0 to 360, 15 to an hour (dKy_getdaytime_hour divides by 15); dKy_daynight_check calls 6:00 to 17:59 day. The environment's setup sets the advance mTimeAdv to 0.02, which dKy_custom_timeset can change. Each setDaytime call adds it only when no event-night stop is set, the save's get-item flag (2, 0) is set, no event runs, and (USA and PAL only) Link isn't on stage "sea" room 14 with stage layer 2 or 3. In rooms that pass time (getTimePass) the clock wraps from 360 to 0 and the day of the week goes up by one; in other rooms it advances only by day before 165 (11:00) and by night from 270 (18:00) to 345 (23:00). At 0.02 a day is 18,000 setDaytime calls (computed).Code
WindIn the USA build, dKyw_wind_set, when no actor has set a wind vector override (mpWindVecOverride), takes the wind's strength from the room's file list (dStage_FileList_dt_GlobalWindLevel): level 0 gives 0.3, level 1 gives 0.6, level 2 gives 0.9, any other level 0. A custom wind power above 0 replaces it, an event wind setting of 2 makes it 0, and it is capped at 1. While mInitAnimTimer is nonzero the wind vector and power jump to the new values; otherwise each call moves each vector component a tenth of the way (at most 2, at least 0.001) and the power a tenth of the way (at most 1, at least 0.005). Since the boat, in the decompilers' not-yet-matching C++ for d_a_ship (procSteerMove), divides the power by 0.5 and caps it at 1, a level-0 room tops out at 0.6 of the boat's 55, 33 units an update, and levels 1 and 2 at the full 55 (computed, once the power has settled).Code
In the USA build, the wind-direction picker dOperate_wind_c (which the atlas reads as the Wind's Requiem's, from its TAKT_WIND sounds; "takt" is the baton) sets mDeg on opening from the current wind vector's yaw and then, with the stick pushed to at least 0.9, sets mDeg to one of eight compass values, 0 to 315 in 45-degree steps. A confirms it: windSet passes dKyw_tact_wind_set a pitch of 0 and a yaw of 180° − mDeg (0 gives −0x8000, 90 gives 0x4000, 270 gives −0x4000). dKyw_tact_wind_set stores both angles for the environment and in the save (dComIfGs_setWindX, setWindY) and sets the tact-wind flag; dKyw_wind_set then builds the wind vector as (cos pitch × cos yaw, sin pitch, cos pitch × sin yaw). dKyw_wind_init reloads the saved angles unless both are −1. B cancels the picker.Code

References

  1. ^ abIn the decompilation's C++ for the boat (d_a_ship, which doesn't yet match the game's bytes as a whole; the USA GZLE01 build), procSteerMove, while the boat isn't airborne (daSFLG_FLY_e off), takes the gap between the wind vector's yaw (cM_atan2s of its x and z) and the boat's facing shape_angle.y, as a 16-bit angle from 0 to 0x8000, and turns it into a wind factor: 1 − gap/0x10000 below 0x4000 (90°), 0.75 − (gap − 0x4000) × 5.493164e-05 up to 0x6000 (135°), then 0.3 − (gap − 0x6000) × 0.00032967035, floored at 0. So the factor is 1 with the wind along the boat's facing, 0.875 at 45°, 0.75 at 90°, 0.3 at 135°, and 0 from 140° to 180° (computed). The target speed is the wind power divided by 0.5, capped at 1, times the factor times 55. The branches then run in this order, one per update. First, on the body animation's frame 7, if speedF is below twice the target (at most 80) it is raised to that and mTargetSpeed is set to the target (10000 if the boat is being moved by force), and nothing else changes speedF that update. Otherwise, while mTargetSpeed is below speedF, firstDecrementShipSpeed lowers speedF toward mTargetSpeed (a tenth of the gap, at most 5, at least 1) and sets mTargetSpeed to 10000 once the gap is under 3; this comes before the sail check, so it applies with the sail flag off too. Otherwise, with the sail-on flag off or a target of 0, decrementShipSpeed eases speedF toward 10 × the stick's value × the cosine of its angle (a twentieth of the gap, at most 0.1, at least 0.015). Otherwise decrementShipSpeed eases it down to a lower target, and cLib_addCalc up to a higher one (a tenth of the gap, at most 0.5, at least 0.1).Codezeldaret/twwrevised
    How we know
    1594    if (!checkStateFlg(daSFLG_FLY_e)) {
1595        cXyz* windVec = dKyw_get_wind_vec();
1596        f32 targetSpeed = *dKyw_get_wind_power() / 0.5f;
1597        int angleDiff = abs((s16)(cM_atan2s(windVec->x, windVec->z) - shape_angle.y));
⋮
1600        if (angleDiff < 0x4000) {
1601            windFactor = 1.0f - angleDiff * (1.0f / 0x10000);
1602        } else if (angleDiff <= 0x6000) {
1603            windFactor = 0.75f - (angleDiff - 0x4000) * 5.493164e-05f;
1604        } else {
1605            windFactor = 0.3f - (angleDiff - 0x6000) * 0.00032967035f;
1606            if (windFactor < 0.0f) {
1607                windFactor = 0.0f;
⋮
    src/d/actor/d_a_ship.cpp lines 1594–1637 at cb24f55 · upstream ↗
    1090void daShip_c::decrementShipSpeed(f32 i_targetSpeed) {
1091    cLib_addCalc(&speedF, i_targetSpeed, 0.05f, 0.1f, 0.015f);
⋮
1095void daShip_c::firstDecrementShipSpeed(f32 i_targetSpeed) {
1096    if (cLib_addCalc(&speedF, i_targetSpeed, 0.1f, 5.0f, 1.0f) < 3.0f) {
1097        mTargetSpeed = 10000.0f;
    src/d/actor/d_a_ship.cpp lines 1090–1098 at cb24f55 · upstream ↗
    22f32 cLib_addCalc(f32* pValue, f32 target, f32 scale, f32 maxStep, f32 minStep) {
23    if (*pValue != target) {
24        f32 step = scale * (target - *pValue);
25        if (step >= minStep || step <= -minStep) {
26            if (step > maxStep) {
27                step = maxStep;
⋮
29            if (step < -maxStep) {
30                step = -maxStep;
⋮
32            *pValue += step;
    src/SSystem/SComponent/c_lib.cpp lines 22–32 at cb24f55 · upstream ↗

    Checked against the evidence by its author, an AI model, 2026-10-22.

    Revised after a check by a second AI model, 2026-10-22.

    In the claim ledger

  2. ^In the decompilation's C++ for the boat (d_a_ship, not yet matching as a whole; the USA build), procSteerMove moves the unnamed angle m0366, which the atlas reads as the tiller, toward mStickMVal × 0x2000 × −sin(mStickMAng) with cLib_addCalcAngleS(…, 4, l_HIO.tiller_speed, 0x100), tiller_speed being 700, and, unless the boat is airborne, passes it to setMoveAngle. Unless the boat is being moved by force (checkForceMove), setMoveAngle sets a rate of 4 − |speedF| / 55 × 3, held between 0.1 and 3.6, blends it toward 3.6 by the unnamed counter m0388 over 15 (3.6 once m0388 reaches 15), and turns the facing by −rate × (m0366 >> 6). With m0388 at 0 and a tiller of 0x2000, that is about 461 units (2.5°) an update at rest and 128 units (0.7°) at speed 55 (computed): the boat turns slower the faster it sails.Codezeldaret/tww
    How we know
    1586    cLib_addCalcAngleS(&m0366, mStickMVal * 0x2000 * -cM_ssin(mStickMAng), 4, l_HIO.tiller_speed, 0x100);
1587    if (!checkStateFlg(daSFLG_FLY_e)) {
1588        setMoveAngle(m0366);
    src/d/actor/d_a_ship.cpp lines 1586–1589 at cb24f55 · upstream ↗
    71const s16 daShip_HIO_c0::tiller_speed = 700;
    src/d/actor/d_a_ship.cpp line 71 at cb24f55 · upstream ↗
    1113void daShip_c::setMoveAngle(s16 moveAngle) {
1114    if (!checkForceMove()) {
1115        s16 initialAngle = shape_angle.y;
1116        f32 turnRate = 4.0f - std::abs(speedF) / 55.0f * 3.0f;
1117        if (turnRate > 3.6f) {
1118            turnRate = 3.6f;
⋮
1120        else if (turnRate < 0.1f) {
1121            turnRate = 0.1f;
⋮
1123        if (m0388 >= 0xF) {
1124            turnRate = 3.6f;
⋮
1126        else {
1127            turnRate += ((3.6f - turnRate) * m0388) / 15.0f;
⋮
    src/d/actor/d_a_ship.cpp lines 1113–1130 at cb24f55 · upstream ↗

    Checked against the evidence by its author, an AI model, 2026-10-22.

    Checked by a second AI model, 2026-10-22.

    In the claim ledger

  3. ^In the decompilation's C++ for the boat (d_a_ship, not yet matching as a whole; the USA build), procSteerMove allows a hop when no event runs, Link isn't barred from control (checkNoControll), the body animation is FN_MAST_ON2 at frame 7 or later, the boat isn't airborne, its forward speed mFwdVel is over 16.5, it isn't being moved by force, and the sailing target speed is at least 11. Then the R trigger's lock (mDoCPd_R_LOCK_TRIGGER) sets the airborne and jump flags and an upward speed equal to the target speed, held between 15 and 40, and sets mJumpRate to (speed.y − 11) × 0.014545455.Codezeldaret/tww
    How we know
    1638        if (!dComIfGp_event_runCheck() && 
1639            !daPy_getPlayerLinkActorClass()->checkNoControll() && 
1640            m0392 == dRes_INDEX_SHIP_BCK_FN_MAST_ON2_e && 
1641            mpBodyAnm->getFrame() >= 7.0f && 
1642            !checkStateFlg(daSFLG_FLY_e) && 
1643            mFwdVel > 16.5f) {
1644            if (!checkForceMove() && targetSpeed >= 11.0f) {
1645                onStateFlg(daSFLG_JUMP_OK_e);
1646                if (mDoCPd_R_LOCK_TRIGGER(0)) {
1647                    onStateFlg((daSHIP_SFLG)(daSFLG_FLY_e | daSFLG_JUMP_e));
1648                    speed.y = targetSpeed;
1649                    if (speed.y < 15.0f) {
⋮
    src/d/actor/d_a_ship.cpp lines 1638–1658 at cb24f55 · upstream ↗

    Checked against the evidence by its author, an AI model, 2026-10-22.

    Checked by a second AI model, 2026-10-22.

    In the claim ledger

  4. ^ abIn the USA build, dKyw_wind_set, when no actor has set a wind vector override (mpWindVecOverride), takes the wind's strength from the room's file list (dStage_FileList_dt_GlobalWindLevel): level 0 gives 0.3, level 1 gives 0.6, level 2 gives 0.9, any other level 0. A custom wind power above 0 replaces it, an event wind setting of 2 makes it 0, and it is capped at 1. While mInitAnimTimer is nonzero the wind vector and power jump to the new values; otherwise each call moves each vector component a tenth of the way (at most 2, at least 0.001) and the power a tenth of the way (at most 1, at least 0.005). Since the boat, in the decompilers' not-yet-matching C++ for d_a_ship (procSteerMove), divides the power by 0.5 and caps it at 1, a level-0 room tops out at 0.6 of the boat's 55, 33 units an update, and levels 1 and 2 at the full 55 (computed, once the power has settled).Codezeldaret/twwrevised
    How we know
    990    if (g_env_light.mWind.mpWindVecOverride != NULL) {
991        wind_vec = *g_env_light.mWind.mpWindVecOverride;
992        strength = g_env_light.mWind.mWindStrengthOverride;
⋮
1022        if (fili_p != NULL) {
1023            wind_level = dStage_FileList_dt_GlobalWindLevel(fili_p);
⋮
1026        switch (wind_level) {
1027        case 0:
1028            strength = 0.3f;
1029            break;
1030        case 1:
1031            strength = 0.6f;
1032            break;
⋮
    src/d/d_kankyo_wether.cpp lines 990–1040 at cb24f55 · upstream ↗
    1062    if (g_env_light.mWind.mCustomWindPower > 0.0f) {
1063        strength = g_env_light.mWind.mCustomWindPower;
⋮
1067    if (g_env_light.mWind.mEvtWindSet == 2) {
1068        strength = 0.0f;
⋮
1071    if (strength > 1.0f) {
1072        strength = 1.0f;
⋮
1075    if (g_env_light.mInitAnimTimer != 0) {
1076        g_env_light.mWind.mWindVec = wind_vec;
1077        g_env_light.mWind.mWindPower = strength;
1078    } else {
1079        f32 scale = 0.1f;
1080        cLib_addCalc(&g_env_light.mWind.mWindVec.x, wind_vec.x, scale, 2.0f, 0.001f);
⋮
    src/d/d_kankyo_wether.cpp lines 1062–1084 at cb24f55 · upstream ↗
    1596        f32 targetSpeed = *dKyw_get_wind_power() / 0.5f;
    src/d/actor/d_a_ship.cpp line 1596 at cb24f55 · upstream ↗

    Checked against the evidence by its author, an AI model, 2026-10-22.

    Revised after a check by a second AI model, 2026-10-22.

    In the claim ledger

  5. ^In the USA build, the wind-direction picker dOperate_wind_c (which the atlas reads as the Wind's Requiem's, from its TAKT_WIND sounds; "takt" is the baton) sets mDeg on opening from the current wind vector's yaw and then, with the stick pushed to at least 0.9, sets mDeg to one of eight compass values, 0 to 315 in 45-degree steps. A confirms it: windSet passes dKyw_tact_wind_set a pitch of 0 and a yaw of 180° − mDeg (0 gives −0x8000, 90 gives 0x4000, 270 gives −0x4000). dKyw_tact_wind_set stores both angles for the environment and in the save (dComIfGs_setWindX, setWindY) and sets the tact-wind flag; dKyw_wind_set then builds the wind vector as (cos pitch × cos yaw, sin pitch, cos pitch × sin yaw). dKyw_wind_init reloads the saved angles unless both are −1. B cancels the picker.Codezeldaret/tww
    How we know
    372    if (stick->getValueStick() >= 0.89999998f && fabs(stick_value) - fabs(field_0x8b8) >= 0.0) {
373        int stick_control = dOw_stickControl(field_0x8d0, stick->getAngleStick());
⋮
377            switch (stick_control) {
378            case 8:
379                if (mDeg != 315) {
380                    field_0x8ca = 7;
381                    mDeg = 315;
382                    mDoAud_seStart(JA_SE_TAKT_WIND_CURSOR);
    src/d/d_operate_wind.cpp lines 370–384 at cb24f55 · upstream ↗
    462void dOperate_wind_c::windSet() {
463    switch (mDeg) {
464    case 315:
465        dKyw_tact_wind_set(0, -0x6000);
466        break;
467    case 0:
468        dKyw_tact_wind_set(0, -0x8000);
469        break;
470    case 45:
471        dKyw_tact_wind_set(0, 0x6000);
472        break;
473    case 90:
⋮
    src/d/d_operate_wind.cpp lines 462–488 at cb24f55 · upstream ↗
    613    cXyz* wind_vec = dKyw_get_wind_vec();
614    s16 temp_r3 = cM_atan2s(wind_vec->x, wind_vec->z) + 0x8000;
615    mDeg = dOw_angleRegular(((180.0f * temp_r3) / (f32)0x8000) - 90.0f);
616    arrowRotate(&field_0xec, mDeg);
    src/d/d_operate_wind.cpp lines 613–616 at cb24f55 · upstream ↗
    661        if (CPad_CHECK_TRIG_A(0)) {
662            windSet();
663            dComIfGp_setOperateWindChangeOff();
664        } else if (CPad_CHECK_TRIG_B(0)) {
665            dComIfGp_setOperateWindCancelOff();
    src/d/d_operate_wind.cpp lines 661–666 at cb24f55 · upstream ↗
    1328void dKyw_tact_wind_set(s16 i_windX, s16 i_windY) {
⋮
1332    env_light.mWind.mTactWindAngleX = i_windX;
1333    env_light.mWind.mTactWindAngleY = i_windY;
1334    dComIfGs_setWindX(i_windX);
1335    dComIfGs_setWindY(i_windY);
⋮
1338    env_light.mWind.mTactWindAngleFlags = 1;
    src/d/d_kankyo_wether.cpp lines 1328–1338 at cb24f55 · upstream ↗
    1010        } else if (g_env_light.mWind.mTactWindAngleFlags != 0) {
1011            tact_wind_x = g_env_light.mWind.mTactWindAngleX;
1012            tact_wind_y = g_env_light.mWind.mTactWindAngleY;
⋮
1018        wind_vec.x = cM_scos(tact_wind_x) * cM_scos(tact_wind_y);
1019        wind_vec.y = cM_ssin(tact_wind_x);
1020        wind_vec.z = cM_scos(tact_wind_x) * cM_ssin(tact_wind_y);
    src/d/d_kankyo_wether.cpp lines 1010–1020 at cb24f55 · upstream ↗
    967    if (dComIfGs_getWindX() == -1 && dComIfGs_getWindY() == -1) {
968        g_env_light.mWind.mTactWindAngleFlags = 0;
⋮
972        g_env_light.mWind.mTactWindAngleFlags = 1;
973        g_env_light.mWind.mTactWindAngleX = dComIfGs_getWindX();
974        g_env_light.mWind.mTactWindAngleY = dComIfGs_getWindY();
    src/d/d_kankyo_wether.cpp lines 967–974 at cb24f55 · upstream ↗

    Checked against the evidence by its author, an AI model, 2026-10-22.

    Checked by a second AI model, 2026-10-22.

    In the claim ledger

  6. ^In the USA build, the time of day is a float from 0 to 360, 15 to an hour (dKy_getdaytime_hour divides by 15); dKy_daynight_check calls 6:00 to 17:59 day. The environment's setup sets the advance mTimeAdv to 0.02, which dKy_custom_timeset can change. Each setDaytime call adds it only when no event-night stop is set, the save's get-item flag (2, 0) is set, no event runs, and (USA and PAL only) Link isn't on stage "sea" room 14 with stage layer 2 or 3. In rooms that pass time (getTimePass) the clock wraps from 360 to 0 and the day of the week goes up by one; in other rooms it advances only by day before 165 (11:00) and by night from 270 (18:00) to 345 (23:00). At 0.02 a day is 18,000 setDaytime calls (computed).Codezeldaret/tww
    How we know
    476    g_env_light.mTimeAdv = 0.02f;
    src/d/d_kankyo.cpp line 476 at cb24f55 · upstream ↗
    502void dScnKy_env_light_c::setDaytime() {
⋮
504#if VERSION > VERSION_DEMO
⋮
506#endif
⋮
510#if VERSION > VERSION_JPN
511    if (strcmp(dComIfGp_getStartStageName(), "sea") == 0 && dComIfGp_roomControl_getStayNo() == 14)
⋮
513        if (dComIfGp_getStartStageLayer() == 2 || dComIfGp_getStartStageLayer() == 3) {
514            var_r30 = true;
⋮
517#endif
⋮
519    if (!dKy_checkEventNightStop() && dComIfGs_isGetItem(2, 0) && dComIfGp_event_runCheck() == FALSE
520#if VERSION > VERSION_DEMO
521        && !var_r30
522#endif
⋮
    src/d/d_kankyo.cpp lines 502–538 at cb24f55 · upstream ↗
    614int dKy_getdaytime_hour() {
615    return dComIfGs_getTime() / 15.0f;
⋮
625int dKy_daynight_check() {
626    s32 hour = dKy_getdaytime_hour();
⋮
628    if (hour >= 6 && hour < 18) {
629        return dKy_TIME_DAY_e;
    src/d/d_kankyo.cpp lines 614–631 at cb24f55 · upstream ↗
    3361void dKy_custom_timeset(f32 i_speed) {
3362    g_env_light.mTimeAdv = i_speed;
    src/d/d_kankyo.cpp lines 3361–3362 at cb24f55 · upstream ↗

    Checked against the evidence by its author, an AI model, 2026-10-22.

    Checked by a second AI model, 2026-10-22.

    In the claim ledger

Sources

Code