Entry 150

The travellers' hum: tobjSeLoopStart and tobjSeLoop

2026-09-26world

Filescrates/piney-stream/src/draw.rsdocs/engine/stream.mdcrates/piney-audio/src/se3d.rscrates/piney-audio/src/driver.rscrates/piney-audio/src/lib.rscrates/piney-world/src/field_ambient.rscrates/piney-world/src/field_world.rscrates/piney-world/src/field_area.rscrates/piney-data/src/field/render.rscrates/piney-game/src/area.rstools/test_sound3d_rs.pytools/test_field_ambient_rs.pydocs/engine/sound.mddocs/engine/field.md

Half of a GAPS field item. Δ's airship and Λ's whale, the servers' travellers that now and then cross a field (TOBJ), hum as they pass. The port drew them in silence.

What the game does

Two functions of main, called only by TOBJ, speak to the synthesizer directly rather than through ccSeOn3DLoop:

• tobjSeLoopStart(pos) (0x0017bf20), from TOBJ::Init for servers 0 and 2. Unless ccSnd +0x133 is already set, it takes the first free loopID slot, keeps it in looptest (0x003789dc), and starts seData[44] (program 31, channel 15, note 60) with a note on at 127 followed at once by expression 0: the loop runs silent. +0x133 is set.
• tobjSeLoop(pos, rate) (0x0017c0d0), from TOBJ::Draw while the traveller's transparency is not 0, sets the loop's pan and volume. The volume is calcVel's arithmetic with a reach of 15,000 per 256 of decay (7,500 here) instead of 5,500, clamped to 0-127 (never -1), then scaled by the transparency, so the hum fades with the picture. Behind the camera it also bends the note as ccSeOn3D does.
• The end. Nothing turns the note off. +0x133 is cleared by initBeforeLoad, so the next ccSndSQLoad (whose sound off silences the note) stops the updates, and by the desktop's ccSndChangeData.
• Ω's sleigh (server 4) calls tobjSeLoop too, but no hum was started for it, so it stays silent.

The port

• piney-audio. se3d::tobj_se_loop_start and se3d::tobj_se_loop build the messages. Driver gains tobj_loop (+0x133, cleared in sq_load and desktop) and looptest. Audio::tobj_se_loop_start and Audio::tobj_se_loop send them on port 0.
• piney-world. Ambient::generate marks tobj_se_start for servers 0 and 2. Tobj::draw emits Op::SeLoop(pos, alpha) before its far test, as Draw calls it. The field runtime sends Op::SeLoopStart once, on the first drawn frame, which comes after the area's ccSndSQLoad as in ccSetupGameCtrl.
• piney-game. The ops become Event::TobjSeLoopStart and Event::TobjSeLoop, heard from the active camera.

A crash found on the way

A 300-frame run of the ambient sweep panicked in cell_height. FIREFLY2::SetBasePosition2 asks the height at its rotated offset (the game's slip). The harness's camera turn had drifted to -8.4 radians, where the game's sine no longer holds (matrix entries near -25), so the offset landed 58,000 away. FIELD::GetHeight wraps only once and read past the map. The game reads whatever memory lies there; the port indexed out of bounds. A real camera's turn is a 16-bit angle, so play stays within -pi to pi. Still, cell_height now gives 0 outside the map instead of panicking, and the harness keeps its turns within -pi to pi.

Checks

• tools/test_sound3d_rs.py's new test_tobj: 1,510 calls of tobjSeLoopStart and tobjSeLoop mixed with ccSeOn3DLoop, random cameras, points and transparencies, run in eemu against the probe: 0 mismatches. test_loop still has 0 in 1,506.
• tools/test_field_ambient_rs.py: TOBJ's calls are now recorded (the starts after Generate, each frame's tobjSeLoop with its place and rate) instead of stubbed. 124 cases of 1,000 frames: 0 mismatches, with 8,463 calls of tobjSeLoop among them.
• the_airship_hums (piney-world): a type 10 field of seed 1 on Δ has the airship; the start is asked; 120 frames each send one hum, the rate rising by 0.01 from 0.01 to 0.8.
• tobj_hum_starts_silent_and_follows_distance_and_fade (piney-audio).

Also: the streams' fog by depth

The streams' models still took one fog value at their centre. They now carry piney_draw::DepthFog like every other model since worklog 0145, so VU1's per-vertex fog is drawn in the streams too (stream.md).