Bug Fixes
- release: share npm web auth across publish batches (natew)
- static: preserve conditional font variants (natew)
Documentation
- universal flat value grammar, per-longhand program merging, value variables, migration setting (natew)
- fold condition-call requirements into the chosen grammar, value variables ride the V3 variables system (natew)
- platform-scope the DOM compiler requirement, single-source skins across frontends (natew)
- RN 0.78-0.87 style capability survey for the v3 grammar (natew)
- safe-area built-in variables, fontVariant test note, RN type-surface check (natew)
- draft the program block encoding, browser-validate its cascade claims (natew)
- design the runtime-compiler boundary and parse cache for dynamic values (natew)
- subject-anchored condition selectors, exit and injection rulings for the program encoding (natew)
- mechanical contract for payload identifier resolution (natew)
- record first parse-cost numbers for the runtime cache model (natew)
- mechanical legacy condition object conversion via shared px-unitless table (natew)
- runtime integration shape, clause modifiers as exact subscription dependencies (natew)
- alpha composition scales, record the normalizeColor.native parity divergence (natew)
- five-lane getSplitStyles wiring decomposition for phase 5 (natew)
- safe-area store landed, variables layer must mount the inset tracker (natew)
- container query table completeness is a config-time check (natew)
- codemod dry-run numbers, transform family is the top migration blocker (natew)
- container queries keep the media key's direction, sizes are the measuring subset (natew)
- RN transform capabilities survey for the transform family design (natew)
- transform family design — individual properties, per-axis variables, fixed native order (natew)
- browser-verify the var-composed translate transition (natew)
- W1+W2 landed, staging limitations recorded (natew)
- the engine contraction — dropping object support shrinks getSplitStyles in v3 (natew)
- W3 landed, W5 running as the engine-contraction test bed (natew)
- theme matching rulings — boundary-prefix rule, accepted scheme-nesting divergence (natew)
- hot-path code rules for the flat value engine (natew)
- scale expands to axis programs, x/y bind space, unitless parity ruling (natew)
- scope the axis-percentage diagnostic to string forwarding (natew)
- merge unit is the clause keyed by modifier set — base overrides keep inherited conditions (natew)
- resolved-program hashing and value immutability are load-bearing invariants (natew)
- native group/container clause evaluation landed through the group context channel (natew)
- engine contraction A/B evidence — 755/769 web, 408/409 native, all gaps shape-only or designed (natew)
- v3 default-on condition conversion and inline program evaluation landed (natew)
- hover capability guard and embedded-token rulings; decision-24 static path proven through the shared pipeline (natew)
- first decision-24 bailout metric baseline — 20.5% over kitchen-sink (natew)
- border family split landed on both contribution paths (natew)
- v3 type performance measured — 1.3-1.8x faster typecheck, static class subset autocompletes (natew)
- correct the type perf record — className autocomplete is structurally unreachable, styled() definition cost regressed 2.47x (natew)
- styled option inference fix lands — v3 now 13% below v2 on instantiations (natew)
- correct the bg autocomplete cause and settle the union cost gap as shape-dependent (natew)
- record that $ is removed from flat candidate syntax, token representation migration still pending (natew)
Chores
- ignore scratch type-perf fixture dirs (natew)