oficina / sheets Public

fix: the review's blocker, and eleven findings under it

§Review by a model that did not write the code returned REQUEST
CHANGES. Discussion 10 carries the verdict; this is the response.

**The blocker: alignment did nothing.** Jspreadsheet writes text-align
as an INLINE style on every data cell, from the column's align or from
defaultColAlign. An inline declaration outranks any selector at any
specificity, so the three toolbar buttons and the Format submenu set
their attribute, reflected aria-pressed, and moved no text. The control
looked like it worked.

The branch's own browser drive did not catch it because it checked bold
and fill — both of which paint properties the library never touches.
That is the third instance tonight of one shape: a probe returning the
same answer whether or not the thing works. The drive now asserts the
COMPUTED alignment, because the attribute was always right.

Two things fell out of fixing it:

- The grid centred everything, while formatpatch.go's comment claimed
  the engine right-aligned numbers and left-aligned text — true of the
  table view and false here. The server now says which cells hold a
  NUMBER (a property of the computed value, which only the engine
  knows) and the stylesheet right-aligns those. A spreadsheet that
  centres its numbers does not look like one.
- defaultColAlign is a WORKSHEET option and was set at spreadsheet
  level, where it is accepted and never read — the exact mirror of the
  handlers, which are spreadsheet-level and ignored inside a worksheet.
  Found by the drive, not by reading.

Also fixed:

- The anchor cursor vanished on filled cells. The rule that marks it
  carries a comment about coming last so the cursor survives; the new
  fill rules land later at the same specificity, so the failure it
  guards against arrived by a new route. It is now an inset ring, which
  is drawn over any fill and costs no layout.
- Delete wrote every selected cell twice. The library already clears a
  selection through its own setValue, which reaches write(); ours was a
  second pass, and Backspace — which the library also handles — went
  once. Inconsistent either way.
- The number format was 128 bytes of free text stored verbatim. An
  Excel pattern admits quoted literals, so that was an arbitrary
  plaintext write into the formats table through a public API. It now
  validates against the five patterns the UI offers, and the UI renders
  from the same list, so the two cannot drift.
- Resize is drag-only and AGENTS.md requires a menu equivalent. The
  Format menu now carries column-width and row-height fields.
- The toolbar's tooltips never rendered: rastrillo's is
  [rst-tip]::after { content: attr(data-tip) }, and only data-tip was
  written.
- A plain viewer got an empty Insert menu; the list row lost its
  thumbnail column below 800px, where rastrillo sets
  grid-template-columns directly rather than through --rst-cols; a
  toolbar dropdown near the right edge overflowed a 320px viewport; a
  colour picked from a toolbar menu dropped focus to <body>.
- Export now sends RFC 6266's two-parameter Content-Disposition. The
  ASCII form drops every non-ASCII rune, so a suite called Oficina
  would have mangled the first Spanish title anybody exported.
- schema.sql was not regenerated for migration 0008.
- The colour-literal guard did not know about oklch(), which this
  branch introduced — a literal oklch would have passed the test whose
  whole job is to catch one.
- Nothing exercised writing an ink: migration 0008's columns and
  applyInk were reachable only by clearing.
- Deterministic export ordering, dead code removed, and three comments
  that had stopped being true.

Two of the review's findings are accepted and NOT fixed here, both
recorded in discussion 10: the read-modify-write in applyFormat is not
under the handle lock, so two concurrent patches to one cell can lose
each other's fields; and a max-size patch is ~12,000 statements in
autocommit. Both want a store change to the locking, which is the most
delicate code in the repo and not a thing to do at the end of a long
session.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0182f7QfojuuqdS4Xd2fYFaL