Entry 141

Chuckie's climb checked: the player is never scaled

2026-10-167mplay

Filescrates/rugrats/src/actors/player.rscrates/rugrats/tests/play.rs

Booting the oracle as another character. The probe harness booted every level with Tommy. The main loop sets the player's character (progress +0x08, 0x800c5680) from fn_80018598(level) before the probe's level hook changes the level. So 0x800c5680 stayed 1.

The fix is a second hook, --onhook 0x800454dc:0x800c5680=2, at the placement loader that builds the player. With it, level 10 starts with Chuckie at (5648, 670, 19228). 670 is his root offset (-354) over the floor.

What 0x800c5680 is. It is the player's own character, so 0x80055904's "not 0x800c5680's character" test leaves the player unscaled whoever he is. Worklog 135 scaled Chuckie's and the others' climbs by +0x4c over -561. That was wrong for the player; it now applies to nobody. The scale belongs to the NPCs' root motion, which the port does not draw.

Checked against the original. In level 10, Chuckie was put at (8945, 670, 12500) by the same seat (template 130) and walked into it. The oracle climbs in table state 13:

661, 652, 639, 622, 605, 502, 398, 282, 154, 25, 6, -14, -22, -17, -12, -23, -34, -50, -72, -95, -114, -133, -152, -172, -193.

The port, unscaled: 660, 651, 638, 621, 605, 501, 397, 281, 153, 25, 5, -15, -22, ... -153. All within 1.

The forward distance differs by up to about 80 in the middle of the climb. Both bodies were put inside the seat, and the original pushes out of it first.

A climb's leftover root y no longer carries on to the ground after it. It had put Chuckie 110 too high on top.

Test: chuckies_climb_moves_as_the_original holds 23 frames within 2.