Four fixes from a review of the runner, each with a regression test that
fails without it:
- A shift inserts cells over the block of every written cell, which moves
that block's columns (or rows) from there on. When a bound input lay in
the way, its data moved and sheared while the run kept the old coordinates
and reported success. The shift is now refused, naming the input and its
place. The output is not counted: it is where the macro writes, and the
second footprint catches data a shift moved into it.
- A dry run or a refused run left a clean workbook dirty: view.zoom,
sheet.activate and the view modes move the saved mark to the current
workbook on a clean document, and the rollback put the workbook back but
not the mark. Saved keeps and restores it.
- overwrite: true answers the question however ask was given; with an
explicit onConflict: ask the answer was ignored and the question came back
forever.
- The dry runs and the run share one step budget as they share one clock;
each phase counted its own 100,000 steps before. ran reports the run's
own steps.
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.
Session::grouped ran its action without a panic guard, so a caller outside
a command that caught the panic found the group still open: Undo and Redo
disabled, the partial edits kept, and the next grouped action nested in the
abandoned group. The group is now rolled back and the panic resumed.
Regression test.
- Workbook::macros is a shared list (Macros, a copy-on-write newtype): a
workbook copy made for a cell edit or an undo snapshot holds the same list,
and the undo budget charges a list once a snapshot owns its own copy.
Before, every mutating command copied every step of every macro, and a
hundred undo entries kept a hundred copies the budget never saw.
- Saving to xlsx drops the last save's items and writes the macros in one
pass each (stash::set_many), instead of a scan of every name per macro.
- A hidden _gc_macro_<x>_<k> name with a key this build did not write is
kept on open and on save, as a visible one is, instead of being dropped
without a warning.
- macros::check is the one check for every path that adds a macro (JSON, a
file, built in code): the model's limits and the xlsx size limit.
Regression tests for each.
The depth check refused any step list at the fifth level, so a fourth Repeat
or If could not even hold an empty body: three levels nested, not the four
MAX_MACRO_DEPTH and the docs state. The check now counts the Repeat or If
steps themselves, in the model's limits and in the parser's pass over the
JSON. The test asks for four and refuses a fifth.
A group open before a command and ended or rolled back inside it lost its
undo step when the command then panicked: the restored history held no entry
for the group's changes, so Undo skipped the state before the group. The
panic path now pushes the outermost such group's step from its start to the
restored workbook. The group stays ended, because its token went with the
command. Regression test for a group ended and one rolled back.
Truncating the undo history to its length at the command's start cannot
bring back an entry that a nested edit.undo popped. The guard now keeps the
undo and redo history as they were and puts both back. The regression test
gains the undo-then-panic case.
The source now carries a note and a column width, and the test checks that
the destination received them while its trust stays Local and no cell is
written.
stash::keys reported a key that get read as missing, because only a later
chunk of it existed. keys now counts chunk 0 and legacy names only, as get
does. Regression assertions in the hostile-names test.
file.open with no file succeeds by queuing a dialog, which execute_headless
turns into an error for callers with no UI; the recording kept the step.
Recorder::truncate forgets it and keeps the byte count right. Regression test.
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.
`_gc_watch_00` parsed as chunk 0, so a hidden one was read as the item's first
chunk and a visible one stopped `_gc_watch_0` from being written. Only the
chunk number as `set` writes it counts now. Regression test.
Paste Link writes a formula into every cell whatever `what` says, so a linked
paste of formats from a file now makes the destination FromFile as a paste of
formulas does. Column widths and notes write no cells, linked or not, and
still carry nothing. Regression test for the three kinds.
Session::grouped ended its group after the action, and when the action had closed the group's document (file.close, or file.open taking the place of an untouched new workbook) the group had gone with it, so the action was reported as failed although it had done what it was asked. A group that is gone because its document is gone now counts as ended; a group ended by someone else while the document is open is still an error.
The history limit is applied once the outermost command has finished, so the lengths its bookkeeping compares stay true (round 3). That left the steps nested commands make unbounded until then: a script of up to 100,000 steps, or a subtotal with a step per group, kept every one of them alive, each with the cells it retains, and before this branch every push trimmed the history at once.
Each nested command now bounds the steps made since the latest mark on its document to the undo limit and byte budget, dropping the oldest of them. The marks are the starts of the running commands, the starts of the open groups and the end of the last group: the positions a command or group collapses its steps by when it finishes, so nothing before them moves. What the bound drops is only what the final trim would drop.
Two boundaries of the group rules were wrong. A command that ended a group open before it was treated as grouped whatever it did next: with an unchanged group (no step) its later edit had no undo entry at all, and with nested edits after the end the entry was the nested command's. The group's step now stays the group's, and the command makes one step of its own, under its name, for what changed after the end, from the workbook the group ended at.
A command that edited and then opened a group it leaves open (a debugger's start) left its nested entries in the history, so one Undo after the group ended brought back a state from inside the command. Such a group now takes the command's start as its own, and its step covers the command's earlier edits.
The record of where the last group ended now holds the workbook and selection it ended at, the history length then and whether it made a step; it lives until the outermost command finishes. A rollback records where it ended too, unless the document is back exactly where the last recorded step left it.
The check ran once per chunk, each a scan of every defined name: an item of many chunks in a workbook with thousands of names compared every pair. One pass now parses each visible workbook-scope name and refuses when it is a chunk of this key below the chunk count. The behaviour and its test are unchanged.
The buffer kept at most 100,000 steps but each step keeps its full parameters, and a file.open step holds the whole file as base64, so a recording of a few opens of a large workbook kept tens of MB alive. Like the journal, the buffer now also stops at 64 MB of steps serialized as JSON. A step that does not fit is counted as dropped, and once a step is dropped every later one is too, so what the buffer holds is always how the recording began, never a sequence with gaps.
The record of the last group step held the document, the history length and a group epoch, and a command reused the group's label when the lengths matched and the epoch was newer. A direct edit after the group (no nested command, so the length still matches) then left the step named after the group alone.
The record now holds the workbook the group's step led to (a weak reference, so it keeps nothing alive). A command keeps the group's name only when the one entry it added starts from the command's own starting workbook and ends at its result, which is provenance by identity rather than by position: an entry a nested command made, a step of an earlier command undone since, and a group step the command edited past all name the step after the command. A later group that rolled back changes nothing, and the rollback no longer needs to clear the record.
- range.setValues with parse: false still turns on Wrap Text for a line break (CodeRabbit).
- Japanese: the fill-handle hint says AutoFill (オートフィル), not paint (CodeRabbit).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Home > Format as Table on a cell or range inside an existing table
failed with "A table can't overlap another table." Excel applies the
picked style to that table instead. The command now returns that table
and its range, as it does for a new table. A range crossing a table's
edge is still rejected.
Without a range, the active area of the selection decides, as it does
for Insert > Table. The native Format menu's Format as Table now runs
home.formatAsTable instead of insert.table.
- automate.trustScripts: only a missing `trusted` defaults to true; a
non-boolean value is an error and leaves the workbook untrusted.
- The refusal message points at Trust Scripts in View > Macros (Alt+F8)
and at automate.trustScripts.
- Catalog row for automate.trustScripts; parity checklist regenerated.
- Script steps run inside an io::ScriptSandbox (thread-local) so
read_file/write_file refuse whatever the parameter is called, and a
step that raises OpenUrl is dropped and stops the script. The id/param
deny list stays as the up-front refusal.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A sheet can hide its outline's level buttons and brackets while keeping
the outline (Excel's Ctrl+8, saved as showOutlineSymbols="0"). It wasn't
modelled, so such sheets showed a wide outline band Excel doesn't.
- model: Sheet::show_outline_symbols (default on, also for older JSON).
- xlsx: read and written on the sheet view.
- engine: view.showOutlineSymbols {on?} (Ctrl+8), undoable.
- ui: no outline bands while it's off.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every string in `values` went through typed input, which reads it with en-US
rules, and nothing said so:
"05.04.2026" (dd.mm.yyyy) → the date 4 May 2026
"02/11/2026" (dd/mm/yyyy) → the date 11 February 2026
"2.500" (thousands) → the number 2.5
"4.25 %" (text) → 0.0425, formatted 0.00%
A workbook built from Turkish-format exports therefore held different values
than the script put in it, and the only way to notice was to read the saved
file back cell by cell.
`parse: false` stores every string as the text it is. A formula is still a
formula — `=` means a formula in a cell, and the ticket asks for that — and
numbers, booleans and null are unaffected either way. A leading apostrophe is
part of the text rather than the marker that makes it text, since nothing is
being marked.
The default is unchanged: `parse` is absent in every existing call, and those
calls behave exactly as before. A `parse` that is neither true nor false is a
usage error rather than a quiet default.
The test pins both sides — the swapped date, the decimal point and the
percentage under today's default; each of the six strings stored verbatim with
`parse: false`, with `=1+1` still calculating 2; numbers and booleans either
way; and the refused `parse` values. Without the change the first string comes
back as a date.
`col("Amount")` read the header through the cell's display text, which is the formula while
Show Formulas is on, so a header such as ="Amo"&"unt" matched nothing. The header is now the
cell's value as formatted, whatever the sheet shows.
A visible workbook-scope name such as `_gc_macro_0_0` matched the storage key and was read as a
macro and removed when the workbook opened. Macro storage is written hidden, so only hidden names
are read back; a person's own name of that shape stays a name.
`Name:` read a name only up to the first period, so `Name:Sales.Q1` looked up `Sales` and
treated `Q1` as a method, and a workbook name called Sales.Q1 could not be used in a macro,
alone or as `Name:Sales.Q1.cell(1,1)`.
The name now runs through periods, and the longest defined name or table that exists and ends
at a period (or the end) is used; what follows is read as methods. So `Name:Data.row` is a
name called Data.row, and `Name:Sales.offset(1,0)` still moves Sales. When no name matches,
the error names the text before the first method, and a bad method after a dotted name is
still refused. Tests cover direct resolution, reference parameters, `$ref` and `{{...}}`,
with and without a chained method.
Opening a workbook gave each clashing macro name a suffix by trying " (2)", " (3)"... in turn,
and each try scanned every macro accepted so far. A file with 1,000 macros of the same
255-character name made about 300 million name comparisons.
Load now keeps a set of the names in use (ASCII case-insensitive, as find compares them) and,
for each shortened name and suffix width, the next suffix to try, so each suffix is looked up
about once. The result is unchanged: the smallest free suffix, within the name length limit,
and the renamed macro still passes the final checks. unique_name keeps its signature. A test
opens, saves and reopens 1,000 same-name macros in JSON and xlsx and keeps them all, and
another counts lookups instead of timing them.
Approvals were stored as (document, FNV-1a hash of the macro's JSON).
FNV-1a is not collision resistant, so a crafted workbook could hold a
harmless macro and a different one with the same hash, and approving
the first would let the second run without asking.
The key is now the macro's canonical JSON text itself (object keys
sorted), so two different macros never share an approval.
macros::hash is replaced by macros::fingerprint.
col("...") always matched against the first row of the range, whatever
its header flag. On an input declared without a header row,
Input.col("Amount") silently picked the first column whose first data
cell read "Amount". After .data, the natural Input.data.col("Amount")
failed, because the first row was then a data row. The header row is
now carried through data, and col("...") is refused with a message when
there is none (an input without a header, a Name: base, or after any
other method).
Opening a workbook accepted a macro whose format version `v` was above MACRO_VERSION. Serde
dropped the fields this build doesn't know, and the next save wrote the reduced macro back
with `v: 2`, so a newer build would read it as a complete version 2 macro.
A version above MACRO_VERSION is now refused by Macro::check_limits, and parse_macro checks
the raw `v` before reading the steps. Such a macro takes the existing per-macro path: it is
left out with the usual single warning (which now names a newer format), and the rest of the
workbook opens. Tests cover JSON and xlsx hidden names, with current and missing versions as
controls.
The policy denies window.* commands, but the View tab's window commands
are view.*: a macro could run view.newWindow, which adds a window that
the run's rollback doesn't remove, or view.hideWindow and
view.unhideWindow, which change the session's hidden windows.
The policy now denies those three. Arrange All changes nothing in the
engine and Switch Windows only lists windows, so both stay allowed.
is_token looked only at the leading identifier, so on a workbook with a
sheet called Output (or Row, Column, or the name of a declared input),
"Output!A1:B2" was read as the Output token, and the step failed with
"the macro has no output anchor" or "unexpected '!A1:B2'". This hit
reference keys, $ref and {{...}} alike, including references the
recorder writes for other sheets. A base followed by '!' is now a sheet
name, and the text resolves as a plain reference.
footprint always ran its dry run with the default 60 second limit, so
macro.run with timeLimit 5 could run for over two minutes (a dry run,
then another after a shift) and fail with "ran longer than 60 seconds".
footprint now takes the run's clock and limit. make_way starts the
run's clock before the dry runs, so the checks and the run share one
limit, and macro.check takes timeLimit too.
Workbook::macros was read straight into typed macros, so one entry this
build couldn't read (a step type from a later build, an unknown
onConflict value, an entry that isn't an object) failed the whole file
with "not a GridCraft JSON workbook". The field is now read as raw
entries into Workbook::macros_read, and macros::load reads each one
with parse_macro, as for xlsx: short-form steps are accepted, the depth
limit is checked before deserializing, and a bad entry is dropped with
the same warning. Only the first 1,000 entries are read, as in xlsx.
The footprint dry run stops at an approved file step and at Solver, but
make_way took a partial footprint with no conflicts as proof that the
run was safe, also after a shift. A macro that exported a CSV and then
set B1 ran under onConflict fail or ask and overwrote B1 without asking.
When the footprint is partial and the rule is not overwrite, the run's
machine now checks every step as it runs: a step that writes over
cells, sheets or objects that held data when the run started stops the
run and rolls it back, with a confirmation under ask and an error
otherwise. make_way now returns the machine for the run along with
whether it shifted.
prxnvt