oficina / sheets Public

A gate that crosses from computed value to painted output

docs-e6 shipped a version of this bug and described the shape, which
turned out to be sitting unnamed in this spec's formatting layer.

Ruling 2 put one engine behind cell values. But a value is not what a
cell shows: 123.456 under a currency format displays as a formatted
string, and if the browser renders that from a raw number then number
formatting is a second implementation of the same truth — the exact bug
class ruling 2 exists to prevent, one layer down. So Go formats and the
client paints strings it is handed. Formats are still edited as
structured records; the client never sends a rendered string.

The gate matters as much as the rule. Gate 4 proves the Go engine matches
LibreOffice; gate 7 proves the client runs; neither proves that what a
person sees is what the engine computed. Docs shipped precisely that: a
conformance test comparing two runtimes of the same module, proving
determinism, which WASM guarantees anyway, and never proving the thing
its rule was about. Gate 11 reads the text actually painted in cells and
compares it against the golden corpus. The rule to keep is that at least
one gate must cross layers, because a suite of gates that all live inside
one layer is a suite that agrees with itself.

Also marks design system v2 as a dated dependency rather than an
assumption, listing shipped versus staged explicitly. Pair, Allocate,
rastrillo doctor and the bare-attribute grammar are ruled but not
released, and neither consuming session could tell which was which by
reading the upstream spec.

Reviewed-By: Fable (dispatched; see spec status line)
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01APtNStXDZ5H2fAc1ZuGqtS