The recorder expanded an automate.runScript step from the workbook as it
stood when the recording stopped, so a script edited or deleted during the
recording changed the saved macro, or dropped the step. The run now leaves
the steps it ran with the recorder (Recorder::ran, by step index, within
the byte budget), and the conversion reads those, falling back to the
workbook only when they could not be kept.
A macro whose format version is above this build's was dropped on open,
so the next save removed it from the file, and the warning said only
that it couldn't be opened. It is now kept as the file held it
(Workbook::macros_unread, shared between snapshots), never read, listed
or run, and written back unchanged by JSON and xlsx saves, numbered
after the readable macros and around visible names of the storage
shape. Kept macros count toward the 1,000-macro limit after the
readable ones; one too large for an xlsx item is left out. The warning
says it is kept and saved back; damaged and over-limit macros keep
their own warning.
check_limits capped the name, description, notes, step count, depth and
input names, and left the shortcut, permissions, output anchor, default
range, column types, command ids, Repeat ranges, If cells and values and
Stop messages bounded only by the item size; the input declaration was
stored unchecked (rows [5, 2], a cell input with cols [2, 5]).
Now each of them has a limit (docs/macros.md lists them), one-line
fields refuse control characters, and a command id is letters, digits,
. and _ (every registered id passes, tested). An If value may be null
(Empty and NotEmpty ignore it); Repeat.max stays the runner's to bound.
An input's least rows and columns can't pass its most, a cell input is
one row and one column with no header and at most one column type, and
no input lists more column types than columns. check_limits takes every
struct apart field by field, so a new field doesn't compile there until
its limit is decided. A macro from a file that breaks a rule is left
out with the usual warning, never changed to fit.
with_names numbered macros 0..N, and stash::set_many refuses to write a
hidden name beside a visible one of the same name, so a visible
_gc_macro_0_0 (from a file, or defined after the file opened, when the
stored names are no longer in the workbook) made every xlsx save of a
workbook with macros fail. The index is GridCraft's choice: macros now
take the free indices from 0 up, skipping every key a visible
workbook-scope name of the storage shape uses (compared as the stash
reads names), in their order. No arithmetic depends on a number from
the file.
The recorder left out every call that opened a dialog, so a script with an
edit before a menu-style step lost that edit from the recording. A call is
left out only when it opened a dialog and changed nothing (the dialog's
confirming call is the step); one that changed the workbook is kept.
Run from inside another command, automate.stopAction rolled back that
command's own group, pulling the workbook from under the rest of its run; and
the recording buffer kept it, so a recording could replay the stop inside the
run it was part of. Nested, the command is refused and the group stays; the
recorder leaves it out (record::NOT_RECORDED).
stash::items and stash::keys listed the first 10,000 items and let the rest
vanish without a word. They return a Listing { found, more } now, so a caller
can say what it did not list.
It keeps a step from the group's start to the restored workbook, so Undo
still returns to the state before the group; the note said the history came
back exactly.
The guard refused any name of the stash's shape, so a visible name a person
had defined, such as _gc_rate_0, could not be edited or deleted from Name
Manager, which cannot pass force. It now refuses a name only when the
workbook holds a hidden workbook-scope item under it; a visible name of the
shape is the person's own, as the stash itself treats it. The test covers
such a name. The query comment says what the copy skip relies on: a query
runs nothing that changes the history.
`Trust` defaulted to `Local`, and `DocState::new` set it, so a
document was trusted unless its maker remembered to mark it. Every
current path that reads a file goes through `file.open`, which marks
it `FromFile`, but a later way of building a document from bytes
(recovery, templates, a new import) would have come out trusted by
omission.
`DocState::new` now takes the `Trust`, and `Trust` has no default.
New workbooks and samples pass `Local`, `file.open` passes
`FromFile`, a new window passes the trust of its workbook, and the
tests that build documents from bytes pass `FromFile`.
`formulas.defineName {"name": "_gc_macro_1_0"}` replaced the hidden
chunk of a stored item with a visible name: the item vanished from
`stash::get` and `stash::keys`, and its other chunks stayed behind as
orphan hidden names. `formulas.deleteName` could remove chunks the
same way, and `formulas.createFromSelection` with `replace` could
replace one from a cell label. `stash::set` refused a clash with a
visible name, but nothing guarded the other direction.
Defining, renaming or deleting a name that `stash::item_key`
recognises (`_gc_<key>_<n>` and legacy `_sc_<key>`) is now refused
with an error that names `force: true`, which lets the call go ahead.
Create from Selection lists such labels in `skipped` unless `force`
is true. The params docs of the three commands say so.
Each `stash::get` looks at every defined name, so listing the items
with `keys` and then reading each with `get` took time quadratic in
the number of names. A file with 20,000 hidden `_gc_macro_<i>_0`
names took about ten seconds in a debug build, and the macro list
reads its items this way from data that comes from the file.
`stash::items(wb, prefix)` now groups the names by item in one pass
and returns each item's text by key, and `keys` lists from the same
pass. Both return at most `MAX_ITEMS` (10,000) items, the first keys
in sorted order, so a file with any number of hidden names can't make
a listing unbounded. A test counts the names each listing looks at.
The xlsx writer drops characters XML 1.0 can't carry: controls other
than tab, line feed and carriage return, and U+FFFE and U+FFFF. A
stash item holding one lost it on save without a warning, and so did
a JSON item, since serde_json leaves U+FFFE and U+FFFF unescaped: a
stored macro parameter changed silently.
`chunk_formulas` now writes each such character, and the marker
itself, as the noncharacter U+FDD0 followed by four hex digits, and
`get` turns them back for `_gc_` items. A marker that isn't followed
by four hex digits naming a character is read as it is. An escape
counts as five units against the 255-unit literal limit and is never
split across literals, and the guard on the stored size allows the
longest item even when every character is escaped. Legacy `_sc_`
items are read unchanged.
A menu-style call with no parameters, such as `file.open`,
`data.fromTextCsv` or `home.formatCells` with `{}`, only queues its
dialog and succeeds. The recorder kept it as a step with empty
parameters, and then kept the dialog's confirming call as well, so
playing the recording back ran the command twice, or failed on the
empty step when played back headless. Only `execute_headless` took
the step back, and only for the open and save-as dialogs.
`execute` now compares the queued UI requests before and after a
top-level command and does not record one that queued a dialog. The
special case in `execute_headless` goes (it keeps turning a missing
file into an error), and with it `Recorder::truncate`, which had no
other caller.
A command that opened an undo group and then returned an error (a `?`
between `group_begin` and `group_end`) left the group open for good:
Undo and Redo refused with "finish or stop the running action first",
later edits made no undo steps, and nothing could stop the group.
`execute` now rolls back every group a failed command opened, as
`group_rollback` does, restoring the workbook, selection, undo and
redo history to where each group began. Only a command that succeeds
keeps a group it opened, which is what a debugger's start needs.
`Session::abort_groups(doc)` and the menu-less command
`automate.stopAction` roll back the groups left open on a document,
for a debugger the person abandons or a caller that dropped its
token. The command is not undoable and works while a group is open,
and the Undo refusal now names it.
automate.allScripts reports trusted since #2178, and scripts reopened from
a file are not trusted yet, so the side-by-side test expects the flag and
compares names and steps after the round trip. cargo xtask parity gives
276 / 292 now (upstream added a command), so the ROADMAP and ui-parity
summaries say so, and this branch's revision rows (275 / 291 to 276 / 292)
sit at the top of their tables with the date they land.
Six fixes from a review of the recorder, each with a regression test (four
fail against the old recorder; the other two bound work that grew with the
input):
- A step that names another sheet with its sheet parameter (cell.set and
about fifty other commands) was tokenized as if on the input's sheet, so a
recorded cell.set {sheet: Two, cell: C4} ran on Two at the input's
relative place. Such a step keeps its references, like a sheet-qualified
one.
- Inlining recorded runs of other macros and scripts stops at the macro's
step limit and counts the rest in dropped, instead of expanding every run
first (a thousand runs of a long macro made millions of steps).
- macro.record {cancel: true} ends a recording without saving it: the way
out of a recording whose macro is too large for the workbook, which could
neither be saved nor ended.
- Formula tokenization reads each reference in the formula and matches it,
instead of listing every column and row of the example (an example as
large as a sheet made a million candidates per formula).
- A selection of several areas keeps its active cell as it is; a token for
the active cell alone put it outside the areas on playback.
- Steps recorded in another workbook are counted in dropped.
Six fixes from a review of the resolver, each with a regression test that
fails without it:
- A bare Row or Column under a general reference key (to, from, at, area,
source, target, destination, location) stays text: Change Spelling's
replacement word and a pivot field's area were being rewritten into the
current row's A1 range. Row.col(2), {{Row}} and {"$ref": "Row"} still
resolve there, and the bare word still resolves under cell and range keys.
- {{{{ in a string is a literal {{, so text that needs the braces can be
written; an unknown token is still an error.
- Input.data on a range that is only a header row, and Name: of a table
with no data rows, are refused instead of reading the header.
- below(0), below(-3) and right(0) are refused like resize(0, 1) is, instead
of becoming 1.
- A defined name on a sheet whose name holds a dollar sign resolves: the
dollar anchors are stripped from the reference part only.
- quote_sheet is the engine's one rule (gridcraft_core::quote_sheet), which
quotes a sheet named like a cell reference, an R1C1 token or a boolean and
leaves a period unquoted.
docs/macros.md says which keys take a bare token and how to write a literal
brace.
A macro could declare an input called Q1, AB12 or Sum1, and a step written
with plain A1 text, as a recorded one is, would then resolve to the input's
range instead of the cell. check_limits refuses such names; the docs say so.
Regression test.
The architecture note named formulas and validation only; it now lists the
Paste Special kinds and linked pastes that write formulas, and the kinds that
carry nothing.
The inputs paragraph described how a run binds inputs and the message it
shows, which the runner adds in a later pull request. It now says what the
declaration is for and that this version stores and checks it.
The panic guard restored the workbook and selection but not the history, so
a nested edit made before the panic left a no-op undo step and the redo
history it had cleared stayed lost. The guard now truncates the steps made
during the command and puts the redo history back. Regression test, and the
architecture note says what the guard restores.
The header and the revision row said 2026-10-11, older than the row above them. The document trust paragraph now says how DocState::trust (provenance) differs from DocState::scripts_trusted (per-session consent for stored scripts, #2178).
A string a catalog lacks shows in English. Coverage, key-parity and
per-language count tests and tests of one language's exact wording are
removed; UI wiring tests take their expected text from the catalog.
cargo xtask i18n-report lists the missing strings per language and
always passes; cargo xtask ci runs it first.
* CI: dev builds keep line tables only
* CI: one integration-test binary per crate
* CI: xtask runs tests with nextest, doctests apart, --keep-going on clippy and wasm
cargo xtask test runs the workspace tests (--lib --bins --tests) with cargo-nextest when it is installed, else cargo test, then the doctests. cargo xtask ci uses the same two steps.
* CI: docs name cargo xtask test
* CI: agent guides name the moved fuzz test
* CI: tests added on main join their crate's test binary
With row 1 frozen, Page Up could land on A1, which is always on screen, while the scrolled rows
stayed a page down (row 10, Page Down, Page Up, Page Up ends on row 1 with rows 2-9 hidden). The
same held for Ctrl+Up into frozen rows and Ctrl+Left / Alt+Page Up into frozen columns.
Arrow and paging keys now record the cell they moved from (`GridState::reveal_from`). When such a
move goes from the scrolled rows (columns) into the frozen ones, the scrolled area goes back to
its top (left). Other reveals (clicks, Find, Go To, Name Box, Undo, the control channel, formula
pointing) leave it unset and keep the view where the user scrolled it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
prxnvt