Official Codebase of 3DCodeVerse
Python
Commits on main
docs: why coupling support did not move the sweep gate — the overlaps are real
20581aa ziyao committed 2 months ago
bestofk: derive k from the data — my first k was too small, in the harness's favour
97c9382 ziyao committed 2 months ago
bench: the compute-matched best-of-k baseline, and the script that reads it
0140662 ziyao committed 2 months ago
cli: flush the summary before exiting — the real cause of "driver output lost"
c411740 ziyao committed 2 months ago
placement: all five "missing planned contents" were the wrapper, not two of them
6f5acf1 ziyao committed 2 months ago
a harness failure is never an agent repair, and a wrapper group is not a missing asset
55c9eb4 ziyao committed 2 months ago
review round 4: the sweep's thresholds, the two probes' depths, and two scripts that lied
78dd379 ziyao committed 2 months ago
textures: a passive offer loses to the recipe — say the ground is what the pack is for
00c6daf ziyao committed 2 months ago
RUNBOOK: a battery launched with 3dcv from a worktree runs the MAIN checkout's code
2a27c97 ziyao committed 2 months ago
RUNBOOK: a third load-sensitive test, and why none of the three gets loosened
8fc42ec ziyao committed 2 months ago
PAPER_WRITING: the scene track's recorded numbers are not usable, and CV3D_SCENE_TEXTURES
12dfccf ziyao committed 2 months ago
docs: D54 carries the measured consequence — fog moved the camera, and that is what made the frame dark
5764b9d ziyao committed 2 months ago
placement: an asset that is scatter AND fog is exempt as instanced
08b400e ziyao committed 2 months ago
probes: a harness_failure finding IS a driver failure, for every caller
5ed5a35 ziyao committed 2 months ago
docs: D57 — the scene texture pack is wired into the loop, behind a switch that is off
454089d ziyao committed 2 months ago
scene: wire the texture pack into the loop, behind CV3D_SCENE_TEXTURES (off)
cca6e63 ziyao committed 2 months ago
"no output" and "the scene did not boot" are different statements
bfedbcc ziyao committed 2 months ago
docs: D56 — a frame rate is a measurement of the renderer that produced it
42e1993 ziyao committed 2 months ago
a frame rate measured on SwiftShader is the box's, and neither gates nor judges
17a5ab1 ziyao committed 2 months ago
bench: re-run the scene placement gate over RECORDED workspaces
26d608d ziyao committed 2 months ago
scene_frames: a lens is never "inside" a god ray — one `nonSolid`, both probes
e6bf41a ziyao committed 2 months ago
docs: D53-D55 and the RUNBOOK rules the first scene battery produced
d686c8f ziyao committed 2 months ago
assembler: name a loaded GLB root after its asset, in the one place that loads them
fd217bb ziyao committed 2 months ago
scene_placement: a mesh that writes no depth is fog, not matter
1b9769a ziyao committed 2 months ago
runtime_js: key the shared-browser endpoint by the runtime_js that spawned it
89f09a6 ziyao committed 2 months ago
render: retry once when the BROWSER dies under a driver, on a browser of our own
6f0019c ziyao committed 2 months ago