`style: {fill: "none"}` (or "auto") went through the colour pass and came out as a solid
fill in the automatic colour. `home.fillColor` reads the same word as no fill, so the rule
style does too now.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
data.validation took any `type`, `operator` or `errorStyle` and fell back
to a default: a typo such as `type: "lst"` saved an "any value" rule
(`type="none"`), which validates nothing, and an unknown operator became
`between`. Each is now an error that lists the accepted values; `any`
stays a type, and the operator aliases (`greaterThan`…) still work.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
data.validation {type: "list", formula1: "Open,In progress,Closed"}, which is
how Excel's Source box takes a list, worked inside GridCraft but was saved as
<formula1>Open,In progress,Closed</formula1>. In a file an inline list is a
string literal, "Open,In progress,Closed" (what Excel writes and what our own
reader expects); unquoted, formula1 reads as a formula of names.
- validation quotes such a source when the rule is created, doubling inner
quotes, using the test list_items already used to tell a literal list from
a reference or a name (now is_literal_list). A source typed with = (a
formula), a reference, a name, a sheet-qualified range and an already quoted
list are stored as before.
- list_items reads a quoted list as a string literal: "" inside stands for
one quote, so an item with a quote reads back the same after the change and
from a file.
Creating the rule is the one place that still knows whether the user typed =,
so the XLSX writer is left alone: it can't tell OFFSET(...) entered as a
formula from a literal list, and a rule read from a file is already quoted.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
home.conditionalFormat deserialized `style` straight into the model's
Style, so a programmatic caller had to write colours as {"Rgb": 16773836}:
"#FFF2CC", which format_range, colorScale and dataBar take, was refused
with a serde message naming internal types. A key the struct doesn't have
(`style: {preset: …}`, `fillColor`) was dropped, which saved a rule with an
empty format that matches and shows nothing. And a cellIs `value` already
written as a string literal ("\"Closed\"", as in the file and openpyxl)
was quoted again into the text "Closed" with its quotes, so it never
matched.
`style` colours (font, fill fg/bg, borders) may now be "#RRGGBB",
{theme, tint} or "auto", and `fill` may be a colour alone (a solid fill);
a fill colour with no pattern is solid. Unknown keys and a non-object
`style` are refused. A cellIs value that is one string literal is kept.
The model's own shape keeps working.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
home.rowHeight and home.columnWidth take pixels at 96 dpi, the model's unit
(header drags and the Row Height dialog send pixels), but their parameter docs
said points. An agent asking for a 30 pt row got `ht="22.5"` in the file, and
the range check accepted 500 while its message said "between 0 and 409 points".
- The params docs say pixels: `height: 20 (pixels) | points: 15`,
`width: 64 (pixels) | chars: 8.43`, and `{width: 64 (pixels)}` for
home.defaultWidth. Pixel values behave as before.
- home.rowHeight takes `points`, Excel's unit (Row Height dialog, `ht` in the
file), as home.columnWidth already takes `chars`.
- Range errors give the limit in the unit passed: 546 pixels (409.5 points),
409.5 points, 1800 pixels, 255 characters. `chars` is checked against 255,
Excel's limit and the widest column a file holds (it went to 256.4 through
the pixel limit).
- The model's size comments say pixels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
view.freezePanes, home.rowHeight, home.columnWidth, the AutoFit, Hide and
Unhide row/column commands and home.defaultWidth always changed the active
sheet: a `sheet` parameter was silently ignored, so a programmatic caller
laying out a sheet that isn't active changed the wrong one, and `Sheet!`
prefixes on `rows`/`cols` were refused as "not a range".
They now take `sheet` (name or index) like the commands that already use
target_sheet, accept a `Sheet!` prefix on `rows`/`cols` (and on Freeze
Panes' `cell`, which was refused), and refuse a prefix that contradicts
`sheet`. The selection belongs to the active sheet, so another sheet needs
explicit lines or a cell. Freeze Top Row, Freeze First Column and Unfreeze
Panes take `sheet` too. Without `sheet`, nothing changes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Business Central's "Open in Excel" exports write booleans as
<c><f>TRUE()</f><v>1</v></c> with no t="b", so the cached value reads back
as the number 1 and the calc oracle reported every such cell as a
`number → bool` mismatch. A cached 1 or 0 now matches GridCraft's TRUE or
FALSE, and the cells are counted apart (`boolAsNumber` in --json, a note in
the text report) so a real number/logical disagreement stays visible.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[@Col] is only Excel's display shorthand. Written verbatim into cell formulas and calculatedColumnFormula, it made Excel repair the file and drop the tables.
Merge & Center and Merge Cells kept only the upper-left value, but
Merge Across left the other values of each row in place, hidden under
the merge and still counted: with 1 2 / 3 4 in A1:B2, SUM(A1:B2) stayed
10 after Merge Across, where Excel gives 4. Each row's merged area now
keeps only its first value, as the other merges do.
home.manageRules took the number of rules once, before deleting, so
{"delete":0,"moveUp":1} with two rules swapped past the end of the
shortened list and panicked (index out of bounds; the session's guard
kept the workbook). The count is now taken after each step.
A formula such as =A1+A2+…+A500 nests one level per operator, and the
evaluator recursed once per level, so past its depth limit (about 400
operators) the result was #CALC!. Excel allows such formulas up to its
8,192-character limit. Chains of arithmetic, comparison and & operators
are now folded in a loop.
A growth series with a step below 1 stopped at once when given a stop
value, since the stop was only checked for rising series. It now bounds
the series in the direction it runs.
