* Add headless GPU backend foundation for composition and flat export (issue #8062 PR 1)
* Fix headless GPU readback bounds, API compliance, and error preservation (#8062)
* Fix GPU readback bounds, context thread spawn error, and test parity (#8062)
* Fix BandRenderer Send bound, PNG reborrows, and clippy lints (#8062)
* Fix effect shadow test construction and remove dead code wrappers (#8062)
* Use short reborrows for Headless renderer without lifetime escaping (#8062)
* Fix clippy collapsible_if and ensure multi-band geometry in flat export tests (#8062)
* Fix automation Rect import, add renderer diagnostics, and harden flat export retries (#8062)
* Expose optional GPU rendering in photocraft-cli (#8062)
* Fix GPU feature test APIs and native target gates
* Fix CLI locked metadata, bounded benchmark params, and protocol diagnostics (#8062)
* Verify unsupported GPU effects before testing band seams
* Fix CLI renderer borrows and GPU-enabled test compilation
* Use compositor buffer ownership API after upstream memory accounting change
* Fix GPU CLI lint failures and validate run environment policy
* Update export renderer test helper for owned compositor buffers
* Fix needless benchmark layer name borrow
* Fix headless GPU backend review findings (#10842)
- release a closed document's GPU caches through a BandRenderer hook
- bound the monitored readback wait and report a stalled GPU as device loss
- reject 32-bit float documents before band readback so exports fall back to CPU
- keep previews at 8-bit RGBA and report the palette/Duotone renderer bypass
- retain renderer diagnostics when encoding fails
- strengthen renderer, save-path and GPU diagnostics tests
* Use valid batch arguments and inspect method in GPU CLI tests
* Add GPU close-hook and 8-bit preview regression tests (#10842)
* Fix GPU CLI review findings (#10912)
- guard the benchmark's modulo divisors for one-pixel dimensions
- count a cold export as GPU only when the renderer context was used
- report every batch file's export warnings, deduplicating only GPU diagnostics
- drain renderer diagnostics before propagating convert and run save errors
- cover renderer use in the GPU precedence, bridge and serve tests
* fix(io): correct renderer binding mutability
* test(cli): verify narrow benchmark inputs and GPU diagnostics
* Fix remaining headless GPU review findings (#10842)
- release a document's GPU caches when file.close, file.closeAll or
file.closeOthers removes it through a command
- honor cancellation while band readback waits, polling in 100 ms slices
- fault the device after a readback timeout so later exports skip it
- report single-layer, CMYK ink-mixing, ORA and PDF CPU paths to a
requested renderer
- keep fallback warnings in their export instead of leaking them into the
next one
- requeue renderer diagnostics when a renderer-aware save's final write
fails
- write MCP/serve diagnostics without panicking on a broken stderr
- document the actual colour-space contract and drop F32 from the GPU
depth list
* Fix remaining headless GPU review findings (#10842)
- release a document's GPU caches when file.close, file.closeAll or
file.closeOthers removes it through a command
- honor cancellation while band readback waits, polling in 100 ms slices
- fault the device after a readback timeout so later exports skip it
- report single-layer, CMYK ink-mixing, ORA and PDF CPU paths to a
requested renderer
- keep fallback warnings in their export instead of leaking them into the
next one
- requeue renderer diagnostics when a renderer-aware save's final write
fails
- write MCP/serve diagnostics without panicking on a broken stderr
- document the actual colour-space contract and drop F32 from the GPU
depth list
* Fix remaining GPU CLI review findings (#10912)
- reject non-UTF-8 PHOTOCRAFT_GPU values instead of silently using the CPU
- release a batch item's document caches before taking the renderer to the
next item, and report fallbacks once
- count warm benchmark runs that fell back to CPU in the summary
- write serve diagnostics without panicking on a broken stderr
- assert that the batch test's --gpu policy reaches the renderer
* Use question mark for cancellable readback errors
* Use question mark for cancellable readback errors
* Keep GPU export diagnostics with their attempt and stop bands on cancel
Encode native .pcraft bytes before draining renderer diagnostics, and
requeue export warnings when a workspace write fails, so callers still
report the adapter or fallback reason with the error.
Move a failed GPU attempt's queued diagnostics into its CPU retry result
instead of leaving them for a later successful export to report.
Stop GPU band readback at the next band after cancellation, and check the
interrupt before each CpuRenderer sink delivery.
Band readback uploads half-precision textures, so reject U16 alongside
F32: their samples cannot survive the upload. Tighten the 8-bit CPU/GPU
parity checks to 1/255.
* Keep GPU export diagnostics with their attempt and stop bands on cancel
Encode native .pcraft bytes before draining renderer diagnostics, and
requeue export warnings when a workspace write fails, so callers still
report the adapter or fallback reason with the error.
Move a failed GPU attempt's queued diagnostics into its CPU retry result
instead of leaving them for a later successful export to report.
Stop GPU band readback at the next band after cancellation, and check the
interrupt before each CpuRenderer sink delivery.
Band readback uploads half-precision textures, so reject U16 alongside
F32: their samples cannot survive the upload. Tighten the 8-bit CPU/GPU
parity checks to 1/255.
* Report partial warm fallbacks and CPU fallback on stderr
The averaged GPU Warm row now says how many warm runs fell back to the
CPU and were excluded from the average, through the existing helper.
The single-layer convert diagnostic test checks that the CPU fallback is
reported on stderr, in GPU-enabled and CPU-only builds alike.
* Keep headless export warnings when a workspace write fails (#10842)
A capability-scoped save encodes the export, then writes it; when the write
fails, the export warnings (folded together with renderer diagnostics) were
pushed back through `BandRenderer::push_diagnostic`, whose default is a no-op.
A custom renderer that overrides only `drain_diagnostics` therefore lost the
adapter or fallback reason that belonged with the error.
Keep them in `Headless` instead and return them from the next
`drain_diagnostics`, alongside the renderer's own queue. The regression uses a
renderer without a `push_diagnostic` override and checks the warnings survive
the failed write and drain exactly once.
Also name the actual source of the depth limit in the headless GPU matrix:
layer content is uploaded as half-precision textures, not band readback.
* Stop CPU bands and CPU retries on cancellation (#10912)
The CPU retry after a retryable GPU error composes through
`FlatSource::bands`, whose shared `emit` sink did not poll the export
interrupt: a cancel was only observed after the whole composite finished.
Check it once per band, as `CpuRenderer` does, for both RGB and CMYK ink
bands. The flat and streamed-PNG retry branches also check before starting
the full CPU composite again, so a cancelled export returns
`IoError::Cancelled` instead of re-rendering the document.
Tests: band-sink cancellation for RGB and CMYK inks, a renderer call counter
proving a second export reaches the renderer, and a pending `wait_slices`
poll observing cancellation set mid-wait.
* Merge origin/main into GPU foundation (#10842)
* Keep GPU-path diagnostics on every failure path, honour cancel in native copy
- save: export warnings of a failed local write stay in Headless pending diagnostics
- render: preview fallback warnings are returned, drained, and added to the doc.render reply
- io: renderer-neutral fallback wording; native single-layer copy checks cancellation per band
- gpu: gate wait_slices tests off wasm32 (#10842)
* Address CodeRabbit review: keep fallback reasons on failed retries, report native-copy and quick-export CPU use, guard diagnostics drain (#10912)
* Accept render diagnostics from the doc.render reply in the serve GPU test (#10912)
---------
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
Conflict in crates/engine/src/jobs.rs: main's save commit point (#5333,
OPEN/COMMITTED/CANCELLED phase) met the batch's cancel state
(CANCEL_REQUESTED/WORK_FINISHED). Both now share one atomic with a
COMMITTED bit; commit vs cancel and the batch's late-cancel guard each
stay one compare-and-swap. A Batch that refuses a repeated or late cancel
reports CancelRefused::Finishing ("finishing") in jobs.cancel's
notCancelled list, still {"cancelled": 0}. control-protocol.md documents both.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Offer TIFF compression, image pyramid and transparency options on Save As and Save a Copy, with matching CLI and automation parameters.
Follow-up: preserve the unsaved state when a copy discards layers over the document's own file, and check the other export formats for the same issue.
Closes#8827.
- A failed input stays in `errors` when `jobs.cancel` or Esc lands while it runs, instead of
showing up as skipped without its reason.
- Esc after the last input is refused: the worker closes the interrupt window once no
checkpoint is left, so `request_batch_interrupt` no longer answers true for a question
that never comes.
- `photocraft-cli batch` without `--format` saves a file PhotoCraft only opens as PSD, the
format `--naming`'s {ext} already plans, instead of asking the encoder for the input's
import-only extension.
- `--recursive` help says the output is flat.
- The app bridge keeps a cancelled Batch's partial result in the error it returns to MCP.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The portable data folder's writability check wrote a fixed, PID-named file with
std::fs::write, which follows an existing symlink and truncates its target.
Create the probe with create_new, try another name on a collision, and remove
only a probe this call created.
AuthorizedWorkspace::write_stream keeps write()'s refusals and atomic replace but
lets the encoder write (and seek back) straight into the temp file, so an
automation PSB save no longer builds the whole file in a Vec.
Resolve conflicts with #5331 (EXR streaming): the block-by-block EXR import now
reads from the same identity-verified handle as the rest of the open path, and
docs/scratch-disks.md keeps both the EXR streaming pointer and the RAM/scratch
admission notes.
Resolve against main's save restructure (Services::save_file/SaveCtl, scratch-integrity checks,
ZipWriter::with_entries, exact encoded_len):
- PSD/PSB saves stream the file model through atomic_write_stream_progress; the length used for
progress and reservations is main's exact encoded_len (the PR's size_bound under-counted lazy
data); export_with/encode_with share checked_stages, so main's scratch checks (previews
included) still refuse unreadable scratch.
- Drop ZipWriter::for_entries in favour of main's with_entries.
- TIFF: take_from_psd moves layer data out of the export's own model (never the document);
body() reserves its exact length, failures are OutOfMemory (not LimitExceeded, which would
trigger the PSB retry).
- Streaming writer: AlreadyExists from the writer is a failed save, not a name collision.
- Remove dead Vec put_widened_rle; channel_data_len is test-only.
Keep the PSD memory admission work and upstream's scratch storage
together. Conflicts resolved:
- Cargo.lock: both sysinfo versions are required. 0.40.1 serves
photocraft-io's adaptive import budget; 0.37.2 serves
photocraft-raster's scratch volume and memory supervision. Cargo
metadata resolved the lock with no other changes.
- crates/engine/src/lib.rs: Session keeps pending_preflights and
next_pending_preflight beside scratch_runtime and scratch_epoch.
Session derives Default, so neither set needs a manual initializer.
Upstream: 655295b2a2
- CLI batch: recursive inputs keep the engine's tree order for {serial};
--format picks the encoder whatever extension --naming writes; the
summary prints before --error-log is written.
- Batch: a jobs.cancel during a write keeps that file in `files`;
Opened Files refuses a default-named result that replaces another open
document's file; a recorded Save leaves a document changed when Batch
writes nothing for it.
- Recording: folded Move-tool nudges add whole-pixel steps; a null
Check Spelling `action` is recorded as Check Spelling.
- Linked files: a document without links lists none missing; a look
whose links changed while it ran is dropped.
- Save for Web: a null or empty destination opens the picker; a rename
while the picker is open writes the picked file.
- Actions: a Check Spelling step fails instead of replacing a dialog
opened by hand.
- Docs: desktop Batch cancel flow, Check Spelling step shape, filter row
history count.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main now takes the manifest envelope from craft-pcraft-format, which has
no full_composite member. The save writes the member after the envelope
when it writes composite/full.png, and read_full_composite reads it back.
With our fix-up 8a0f959d29 on top of the author's head 072a87e7b0.
Conflicts with the batch resolved: Image Size keeps #10702's point sizes and
writes `scaleStyles` only when true (no key and false play the same in the
reference application, so recorded steps save back unchanged); Convert to
Profile takes #10704's options; new tests carry #10743's `encoding` field;
#10702's absolute-intent test expects the new Convert to Profile flags.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* perf: reuse large buffers so a pinned glibc trim threshold costs no speed (#10479)
With MALLOC_TRIM_THRESHOLD_ set (#8743), glibc maps every allocation over
128 KiB fresh, so tiles, compositor bands and filter scratch were page-faulted
in on every call: viewport compose 3.6x slower, Gaussian Blur up to 1.8x,
.pcraft save 2.4x.
- photocraft_raster::scratch: a process-wide, capped pool of large Vec
buffers (u8/u16/f32/[f32; 4]). A miss drops the oldest buffers worth its
size down to a 256 MiB floor, so peak memory stays near glibc's defaults.
The engine releases the pool after any command that closes a document,
Headless::close does too, and an idle thread releases it after 2 s unused.
- Tiles, compose Buffer/masks/bands, algo Image and the blur paths, engine
adjustment/filter scratch, GPU tile upload and io import buffers go through
the pool.
- .pcraft save reuses one zstd FrameCompressor per worker (byte-identical).
- interactive_bench reports minor page faults per run.
36 MP, trim threshold set vs glibc defaults: viewport 5.6 vs 6.1 ms (was
22.7), Blur r4 445 vs 437 ms (was 852), Curves CPU proxy 35.0 vs 35.7 ms
(was 50), flatten 146 vs 147 ms (was 237), save .pcraft 251 vs 230 ms (was
545). RSS after closing a 5400x7200 document: 21 MB with trim set.
* fix(linux): restart once with glibc's trim threshold set so closed documents return memory (#8743)
With glibc's default malloc, the memory of a closed document stays with the
process: freeing a large buffer raises glibc's dynamic mmap threshold, so
later tile buffers land in heap pages glibc keeps. A session that opened and
closed a 5400x7200 document held ~450 MiB afterwards.
photocraft_engine::malloc_tuning::apply(), called first in both binaries'
main, re-executes the binary once with MALLOC_TRIM_THRESHOLD_=131072 on Linux
with glibc: same PID, arguments, argv[0] and environment, in safe Rust
(CommandExt::exec). A malloc variable or glibc.malloc.* tunable the user or a
package already set is left alone, so setting MALLOC_TRIM_THRESHOLD_ opts
out. No-op on macOS, Windows, musl and wasm.
Depends on the scratch-buffer pool (#10479), without which the pinned
threshold slows filters, previews, flatten and .pcraft save by 26-209%.
* fix: corpus tests build again; return remaining pooled buffers (#10479 review)
- Corpus-gated io tests and the Photoshop-oracle test moved `.px` out of a
`Buffer`, which now implements `Drop` (E0509 with `--features corpus`):
use `into_px()`.
- io: masks, the merged image, extra channels, smart-filter masks,
smart-object pixels and patterns give their `interleave` buffers back to
the scratch pool after writing them.
- compose: recycle the mask values of a pass-through group, `effect_shape`
and both `split_parts` callers.
- Styled Free Transform preview copies each band out instead of moving it,
so the band returns to the pool.
- interactive_bench: the fault count covers each whole run, untimed setup
included; say so.
* fix: corpus tests build again; return remaining pooled buffers (#10479 review)
- Corpus-gated io tests and the Photoshop-oracle test moved `.px` out of a
`Buffer`, which now implements `Drop` (E0509 with `--features corpus`):
use `into_px()`.
- io: masks, the merged image, extra channels, smart-filter masks,
smart-object pixels and patterns give their `interleave` buffers back to
the scratch pool after writing them.
- compose: recycle the mask values of a pass-through group, `effect_shape`
and both `split_parts` callers.
- Styled Free Transform preview copies each band out instead of moving it,
so the band returns to the pool.
- interactive_bench: the fault count covers each whole run, untimed setup
included; say so.
* fix: recycle the masks of adjustment-atop and shape layers (#10479 review)
* fix: recycle the masks of adjustment-atop and shape layers (#10479 review)
* fix: recycle effect hiding masks (#10479 review)
* fix: recycle effect hiding masks (#10479 review)
* fix: recycle knockout, CMYK and GPU shape masks; release the pool when Split/Merge Channels close a document (#10479 review)
* fix: recycle knockout, CMYK and GPU shape masks; release the pool when Split/Merge Channels close a document (#10479 review)
* fix: release the pool when a closed document's cancelled workers exit; idle-release switch for tests; don't pool PSD planes (#10479 review)
* fix: release the pool when a closed document's cancelled workers exit; idle-release switch for tests; don't pool PSD planes (#10479 review)
* fix: release the pool for workers cancelled before their document closes; take the pool lock for budget and idle-release changes (#10479 review)
* fix: release the pool for workers cancelled before their document closes; take the pool lock for budget and idle-release changes (#10479 review)
* fix: idle release rechecks activity under the lock; try_take_exact releases the pool and retries before failing (#10479 review)
* fix: idle release rechecks activity under the lock; try_take_exact releases the pool and retries before failing (#10479 review)
* test: cover a successful try_take_exact miss; name the failed-reservation test for what it checks (#10479 review)
* test: cover a successful try_take_exact miss; name the failed-reservation test for what it checks (#10479 review)
* fix: pool the final GPU effect shape buffer and return it after use (#10479 review)
* fix: pool the final GPU effect shape buffer and return it after use (#10479 review)
* docs: describe scratch pool lifecycle regression invariants
* test: verify polling releases cancelled document scratch buffers
* perf: recycle remaining transform selection and PSD scratch buffers
* fix: free PSD gather bands before allocating channel planes
* test: assert exact scratch take reuses pooled storage
The three version 12 fixtures were tracked inside crates/psd/fuzz/corpus,
which crates/psd/fuzz/.gitignore ignores and cargo-fuzz rewrites (cmin) or
deletes; the psd and engine test binaries include_bytes! them, so cleaning
the corpus broke the build. They move to crates/psd/tests/fixtures/atn with
a README (sizes, purpose, sha256; the seeds note folded in) and an
ATTRIBUTION.md row; the fuzzer takes the folder as an extra corpus dir.
photocraft-cli recognised an extensionless Actions file only by version
word 16, so a version 12 set without .atn (classic Mac OS Photoshop 5)
went down the JSON path and failed with a UTF-8 error. It now accepts 12
and 16, the versions photocraft-psd reads. Input: v12-commands.atn copied
to "Legacy Actions" and passed to load_actions_file.
Co-authored-by: yoriswind <164435053+yoriswind@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Read bounded original SVG through the platform clipboard service for native
Paste options and action playback. Preserve a newer internal copy when an
older OS source remains unchanged, and report failed or panicking reads
without losing the existing copy. Retain ordinary raster clipboard routes.
Add source replacement, mode/output, Undo, action playback, stale-source,
size-bound and panic/retry regressions. Completes the app intake for the
source-backed backend adapted from Skyfuller1303's PR #3409.
Validation: 104 unique UI/Actions/native-service passes, one existing manual
clipboard ignore; actual app-model Pixels/Shape renders and saved Path;
2921 engine package passes (22 ignores), adversarial pass, all-target strict
engine/UI/app Clippy, full format, 31-crate layers and standard wasm/HEIF.
The doc.save row, docs/development.md, the MCP doc_save description, the
file.saveACopy params text, the --pcraft-full-composite help and the
option's doc comments said the option writes composite/full.png. A save
leaves it out when the canvas is past 2^26 pixels, when its pixels cannot
be allocated or encoded, or when a ZIP bundle would pass its limits. They
now say so and point to the manifest's full_composite member, which names
the entry when it was written.
Conflicts resolved as unions of both sides: codecs re-exports (PngRows + TiffBands), smart_cmds source imports
(main's source_surface_from_buffer and cache API with #8571's svg_source_image_in/convert_svg_source; main's
duplicate rendering import dropped), io flat.rs test modules (#8536's banded TIFF tests and main's PNG ones),
menus/invoke.rs (#8571's smart SVG intercept and main's layer.selectUnderCursor), state.rs (#8311's
type_box_overflow and main's type_insertion_style), the canvas_geom test keeps the is_finite guard, docs
status lines keep both change lists, i18n catalogs keep both sides' rows (hi.tsv: main's wording where both
added the same string).
UPDATE_APP (native builds; craft-update is a native-only dependency) carries our 128 px icon and accent (#2f7bf5, its colour in ArtCraft Launcher). The updater, the self-update
helper and craft-update's startup prompt all get them through crate::UPDATE_APP; the
prompt draws the icon, the accent and "by ArtCraft" (craft-update's default developer).
craft-update is pinned to storytold/craft-libs main at 1065653e7 (#434).
Clippy fixes for Rust 1.99 in untouched code (same logic):
- apps/photocraft/src/app_dirs.rs
- crates/codecs/src/codecs/via_image.rs
- crates/engine/src/canvas_geom/tests.rs
- crates/ui-egui/src/panels/blend_preview.rs
Main's noop_step now takes the step's params (an Arrange that cannot move the layer is a no-op too); it is public for the CLI's run and batch, which pass them.
- Smart-object contents (Replace Contents, Convert to Layers, Stack Mode,
internal Edit Contents) refused any SVG whose canvas area times its
node count passed 24 MP, e.g. a 1920x1080 layout with 12 shapes.
decode_source now uses the bounded image parse with File > Open's
layered import (import_svg_bounded_images). Convert to Layers' shape
cache cap is 1 GiB in the document's format, the import's own budget,
not 24 MP (four full-canvas shapes in a 4K document failed).
- The bounded resolver failed the whole drawing on an embedded image it
can't read (an unknown MIME type, a corrupt PNG); it now leaves it out
as usvg does, so such SVGs place and render again. Byte and decode
overruns stay errors.
- resvg gives each embedded SVG image a scratch pixmap the size of its
target and frees it after, so the scratch budget counts the deepest
chain of nested SVG images, not every sibling: 12 SVG icons on a 2 MP
canvas, or one at a 12 MP render, no longer fail.
- Main's shared nesting budget (#9397) now covers the bounded path too:
511 + 511 levels across a drawing and its embedded SVG overflowed a
1 MB stack.
- An AI working copy is regenerated through VectorCraft only when
VectorCraft is the editor; Illustrator or another program gets the
original bytes. VectorCraft is recognised by bundle id, a known
install, or its name (vectorcraft[.exe], vectorcraft-<version> AppImage,
ai.storyteller.vectorcraft).
- updateSvg refuses an AI/PDF/EPS source (updateVectorSource saves it
back in its own format); the doc no longer calls updateSvg "existing".
- Edit Contents on contents already open in the internal editor switches
to that document, as main does, instead of an error.
- Stack Mode on an SVG refuses up front when one float frame per layer
would pass 1 GiB (the layered import no longer refuses for it).
- A stored display preview still fails on an embedded image it can't
read (strict budget, its own import_svg_preview_with_work; external-
editor saves and conversions leave the image out as placement does); skipped embedded SVGs still use up the XML byte
budget; an image under an unknown MIME type or with no pixels is left
out uncharged.
- Merge: the PSD import reads a preview's native source from the
document's global blocks; main's owned open moves them out of the
file, so every stored preview was dropped.
- Main's strict Clippy rules on the PR's app code (explicit import, a
Unix-only test helper, one long opt-in test).
Co-authored-by: Mocky Digital <282319807+mockydevs@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>