* Dev profile: line tables only for workspace crates, no debug info for dependencies
* xtask ci: dev-profile gate, nextest, tests without examples, wasm --keep-going
* One integration-test binary per crate
* xtask ci: run effectcraft-gpu's shared-device tests in one cargo test process
* xtask ci: run every stage and report all failures
* xtask ci: name the stage type (clippy type_complexity)
* xtask ci: run every test pass and report all failures
* cargo fmt
The text editor's Select All and Undo arms ignored Alt, so a Deselect All
rebound to Cmd+Alt+A or Cmd+Alt+Z was taken as Select All or Undo while a
text edit was open.
The arms now match exactly the chords the editor owns (Cmd/Ctrl+A, and
Cmd/Ctrl+Z with or without Shift); their Alt variants fall through to the
Deselect All binding match. Chords the editor owns still win over a rebound
Deselect All, so Undo keeps working while typing.
Test: a UI kittest rebinds Deselect All to each Alt variant and checks that
Cmd+A and Cmd+Z still edit the text while the rebound chord ends the edit.
* Use chrono for UTC/calendar conversion instead of hand-rolled copies
Replaces the hand-rolled civil_from_days / days_from_civil copies (log
timestamp, ISO metadata dates, Align Video to Data timestamp parsing, the
Numbers effect's date, the Home and Render Queue date columns, and the
benchmark report date) with chrono as a workspace dependency. Output formats
are unchanged and pinned by the existing tests. AGENTS.md now says time goes
through chrono.
Follow-up to storytold/cadcraft#1785
* Keep 5-digit years unprefixed; derive Numbers weekday from the same date
chrono's %Y writes a '+' before years past 9999, where the old code wrote
{y:04}. timestamp, iso_date, format_utc, utc_stamp and xtask's utc now write
the year with {:04} again and let chrono format the rest, so a far-future
clock keeps the plain zero-padded year.
date_from_days took its weekday from the raw input modulo while the
year/month/day fell back to the epoch for day counts outside chrono's range,
so an unrepresentable count printed a fallback date paired with the input's
weekday. The weekday now comes from the same resolved date, so both agree.
Tests cover the year-10000 timestamp and the out-of-range Numbers date.
* Keep the whole date in format_ago for five-digit years
format_ago cut format_utc to 10 bytes, so a year past 9999 lost its last
digit (10000-01-0). Take the date up to the space instead. Tested.
* Keep Numbers dates exact past chrono's range
date_from_days converted the day count by building a Unix timestamp and
calling DateTime::from_timestamp, which falls back to the epoch (and so
reads 1970-01-01, a Thursday) once the date leaves chrono's NaiveDate
range near year ±262143. Numbers' `value` slider is a plain scalar with
no range guard, so a keyframed count reaches that: Long Date of
100_000_000 showed "Thursday, January 1, 1970" instead of the real
"Monday, September 13, 275790" -- exactly the plausible unrelated-date
fallback the review rejects, and the derived weekday matched the stand-in
date, not the input.
Route it through the 400-year Gregorian cycle like pdfcraft 9d4cb3ca:
chrono converts the day within one cycle (num_days_from_ce of the rebased
rest, always in range) and the whole cycles are added back as years, in
i128 so i64::MIN/MAX can't overflow. chrono stays the only calendar
logic. The weekday still comes from the resolved date, so it always
agrees with it.
Tested: date_from_days_past_chrono_range pins the pre-chrono dates and
weekdays at/over chrono's bound on both sides of the 2000-01-01 epoch
(and that the extremes don't overflow); numbers_formats adds the same
through the public numbers_text entry (Long/Short Date). cargo test
-p effectcraft-effects, fmt --check and clippy -D warnings all clean.
* time: preserve out-of-range years instead of the epoch
`DateTime::from_timestamp(..).unwrap_or_default()` changed the existing output:
inputs the old hand-rolled civil-from-days converted (e.g. `10_000_000_000_000`
s → 318857-05-20 17:46:40) collapsed to the epoch. Restore the old conversion
domain with a cycle-based helper: reduce by whole 400-year Gregorian cycles
(146_097 days), convert the in-range remainder with chrono, and carry the year
back as i128 (saturating to the output type). Only the year leaves chrono's
range; month, day and time of day stay exact. chrono remains the only calendar
logic (NaiveDate::from_num_days_from_ce_opt); no new features.
Sites (UTC):
- crates/engine logging.rs `timestamp` (u64 ms): far-future clock keeps its year.
10e12 s → "318857-05-20T17:46:40.000Z", u64::MAX ms → "584556019-04-03T14:25:51.615Z"
(was "1970-01-01...").
- crates/engine media_browser.rs `iso_date` (u64): 10e12 s → "318857-05-20T17:46:40Z",
u64::MAX → "584554051223-11-09T07:00:15Z".
- crates/ui-egui panels/home.rs `format_utc` (i64): 10e12 → "318857-05-20 17:46",
i64::MIN → "-292277022657-01-27 08:29".
- crates/ui-egui panels/render_queue.rs `utc_stamp` (u64): 10e12 → "318857-05-20
17:46:40", u64::MAX → "584554051223-11-09 07:00:15".
- xtask bench.rs `utc` (u64): 10e12 → "318857-05-20 17:46 UTC", u64::MAX →
"584554051223-11-09 07:00 UTC"; added a test.
* Name the 400-year range extension as the one exception to the time rule
The rule said never to hand-roll calendar math, but the range fix
(shifting by whole 400-year Gregorian cycles so dates past chrono's year
range keep their old output) is cycle arithmetic, so the rule and the code
disagreed. The rule now names that exception: the library still computes
month, day and time, and only whole cycles are carried in the year.
* Say the library computes month and day; time-of-day is plain arithmetic
Narrows the range-extension wording: the date library computes the
calendar fields, and splitting seconds into hours, minutes and seconds
is ordinary arithmetic, which is what the code does.
While a text layer was being edited, Edit > Deselect All and its shortcuts
selected every character (or did nothing) instead of ending the edit and
deselecting the layer.
- deselect_all now commits the open edit first, through the same route as
text.endEdit: a created-but-empty type-tool layer is discarded, anything
typed is kept, and no extra undo step is introduced.
- While the in-canvas editor holds the keyboard, the editor forwards the
command's currently assigned shortcut (and any function-key chord) so
Deselect All reaches the command from the comp; bare-letter bindings stay
typing, and Select All still selects characters.
- The Project-panel focus redirects (invocation and menu enabled-ness) stand
down while a text edit is open, so the menu and shortcut end the edit even
when focus stayed on Project after Layer > New > Text.
Tests: three engine cases (commit + discard semantics), six UI kittest cases
(chords, F2, rebinding, menu enabled-ness, Project focus), RED at base.
* segment: bound checkpoint tensors and pickle work
* segment: bound entry size, memo clones, and total tensor elements
* segment: cap cumulative decompressed storage bytes
Shadow maps use the full scene-derived size (point-light faces capped
at 2048) under a 26M-texel scene budget halved largest-first; the PCF
filter measures its reach in output pixels with a two-texel minimum,
samples texel centres on a jittered adaptive 5..10 grid, and compares
each texel against the receiver plane's own depth so wide kernels never
self-shadow lit casters. The WGSL mirror is parenthesised (Tint rejects
the old form) and guarded by a shader-wide grammar test; the CPU
reference and shader remain step-for-step parity-tested.
Reload Footage probed every item the way File ▸ Import does by default,
so a still such as Photo_2.jpg with Photo_1.jpg beside it was grouped
into a run and measured from the run's first file. The still took
Photo_1's size. Sequence items still probe as a run; everything else
reads its one file.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
A raster footage layer's Timeline menu now also has to leave out Create Shapes from Vector Layer. The viewer check keeps the solid selected, so the menu has to follow the layer under the pointer rather than the selection. With the viewer's caps merge made a no-op, the test fails at the viewer check.
file.projectSettings takes frameStart (0|1), the Project Settings dialog shows Frame Count, and scripts get app.project.framesCountType and displayStartFrame. The setting was already read by the Frames and Feet + Frames time display; nothing could change it.
Double-clicking a property name in Effect Controls did nothing. The test double-clicks the name and checks the Timeline marks the property revealed and shows its row. Without the fix nothing is revealed.
The focused panel's outline was painted before the panel body, so body fills hid it. The test finds the outline in the painted shapes and checks no filled rect painted after it reaches its one-pixel ring. Without the fix a body fill overlaps the ring.
Right-clicking vector footage in the Timeline or the Composition viewer offered no vector conversion. The test imports an SVG and checks the item is absent for a solid, present on the SVG layer in both menus, and converts it. Without the fix the menu item is missing.
Resampling footage audio divided by the requested mix rate, so a rate of 0 panicked. The test decodes a 44.1 kHz WAV at rate 0 and expects a read of the requested length. Without the guard it panics with 'attempt to divide by zero'.
Nothing cleared a `Value::Layer` reference when its layer was deleted, so the id stayed behind:
`env_pixels` resolved it to nothing and rendered no environment at all, while the UI showed "None",
which is documented as "fall back to the comp's Environment Layer". The delete path already
unparents children and drops a track matte that pointed at a deleted layer; a layer reference now
goes the same way, in the same history step, so undo restores the layer and the reference together.
This covers every layer-reference property, not only an Environment light's Source: 73 of them
across the effects (Clone Source, Set Matte, Back Layer, Water Surface, Sky, Ground…). Keyframes
carry the same value kind as the property, as in `essential.rs`, and are fixed too.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The setters go through `comp.settings`, which defaults to the Composition Settings dialog's
centre anchor, so `comp.width = 800` on a 400-wide comp shifted every unparented layer — and
all of its position keys — by 200 px. A script that resized a precomp then found its pixel
coordinates had moved under it.
After Effects' width and height setters only change the frame; moving the contents is what
"resize from centre" scripts do for themselves. The setters now pass `anchor: 0` (top left).
The dialog, which is where the anchor belongs, is untouched.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The Composition Settings dialog draws 「ピクセル縦横比」 and 「フレーム縦横比」, and its checkbox reads
「縦横比を 1:1 (1.00) に固定」. Three existing rows said アスペクト比.
Co-Authored-By: DeepSeek V4.1 Flash <noreply@deepseek.com>
The Timeline drew a layer-valued property (an Environment light's Source, a
Displacement Map's layer parameter) as read-only text holding the raw layer
id, so a light's Source could not be chosen in the UI at all - only through
prop.set from the command line, scripting or MCP. Effect Controls already
had a working drop-down for exactly these values: extract it into one shared
widgets::layer_picker (the current entry, "None" first, then "N. name" for
every layer of the comp, the list built only while the popup is open) and
draw it from both panels.
In the Timeline the pick writes through the same set closure as every other
value arm - an undoable prop.set (layer id, or null for None) - and
registers timeline.prop.<uid>.value like its neighbours, so the row names
the layer instead of exposing its id. Effect Controls keeps its rect,
automation id and write; its closed dropdown's "None", previously a
hardcoded literal, now follows the interface language. The dead "Layer {n}"
id text leaves the i18n catalogs with it.
ui_layer_source_picker drives both panels headlessly (open, pick, undo,
dismiss, current-entry highlight, translated None, picking None to clear a
set value), and an ignored wgpu snapshot test keeps the visual record
reproducible.
Every selector of a comma list cloned the whole declaration block, and every
element scanned every rule, so a sub-megabyte <style> could demand billions of
string pairs. Selectors now index one shared block, the sheets are capped at
20 000 selectors and 50 000 declarations, rules are bucketed by id/class/type
so an element checks only candidates, and matching work over all elements is
charged against a 50M-step budget that fails the import instead of doing it.
Found while testing crates/geom/src/lib.rs:362-390 (pivot check at line 367)
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The search test now pins every row (layer, Contents, group, path/fill/
Transform) so a future regression that re-adds the inner Contents row or
reorders the flattening fails loudly. Docstrings cover the functions the
diff touches (group_row, all, rec, push_group) for CodeRabbit's
docstring-coverage pre-merge check.
- scriptui.js: replace the self-comparison NaN check (o.max === o.max) with
the explicit Number.isNaN(o.max) (S1764).
- sw.js: mark the fire-and-forget cache.put() with void and a .catch() so the
promise rejection is handled (S9383).
- prelude.js: return __err(...) from the activeCamera getter so every path
yields a value (S4275); __err always throws, so behavior is unchanged.
The draft records the project epoch and comp it was opened for. A script or the
control channel opening a project or activating another comp while it is open
closed nothing before, and OK added the primitive (and switched to Advanced 3D)
in whatever comp was active then, even in a new project reusing the comp id.
aiongg