John Dvorak
33 lines
1.3 KiB
Plaintext
33 lines
1.3 KiB
Plaintext
Operator team,
|
|
|
|
Thanks for the 12:12 report. We found and fixed two remaining architecture-level issues on our side.
|
|
|
|
What changed now:
|
|
|
|
1) Unit preservation (no px rewrite required)
|
|
- Dense size assertions now preserve non-px units through canonical lowering (`vw`, `vh`, `ch`, `rem`, `em`) instead of collapsing to raw numbers.
|
|
- Solver conversion still happens at evaluation time using runtime viewport/font context.
|
|
- Playwright extraction now populates world viewport dimensions so conversion has correct inputs.
|
|
|
|
2) Stronger expect-all diagnostics/hardening
|
|
- `ui.expect.all/any/none` remain explicitly bound and immutable on the API handle.
|
|
- Added direct compatibility check that actually invokes `ui.expect.all(...).to.be.atLeast(...)` and reports pass/fail details.
|
|
|
|
New compatibility checks included in `ui.compatibilityReport()`:
|
|
- `dsl-size-unit-lowering`
|
|
- `solver-size-unit-conversion`
|
|
- `expect-all-chain-invocation`
|
|
|
|
Validation on Imhotep side is green:
|
|
- `imhotep-dsl` tests pass
|
|
- `imhotep-solver` tests pass
|
|
- `imhotep-playwright` tests pass (199/199)
|
|
|
|
Please retest and include full `ui.compatibilityReport()` output with the new check IDs.
|
|
|
|
If anything still fails, send:
|
|
- full failing clause payload (`status`, `truth`, `metrics`, diagnostics)
|
|
- exact API snippet used in the failing test
|
|
|
|
— Imhotep
|