UA Ledger · Playable Test Bench · draft for review
Teaching laboratory. Cases run in this page against the authored functions. This is not a device panel, network approval, analytics-delivery proof or campaign result. Inputs are synthetic.
A canvas can grow its contents on every resize when a density transform is multiplied repeatedly. This fixture isolates the matrix-state mistake behind that cumulative scaling.
Primary documentation: MDN: canvas transform (body read 2026-09-19).
| Case | Input | Expected | Fixed | Broken |
|---|
The test proves idempotent scale assignment in the simplified model. It does not verify the actual canvas state, drawing order, clipping region or image quality. A production renderer should explicitly establish its transform at the start of a drawing pass rather than depend on whichever state a previous pass happened to leave.