Maintenance keeps the existing installation context, and a major upgrade of an
existing per-machine installation without ALLUSERS=1 is rejected rather than
installed per-user.
The policy check now reads the compiled AppSearch and RegLocator rows for the
remembered-folder properties (registry root, key, value name and type for both
scopes) and runs the DIR, FOLDER, PATH and legacy exe actions in sequence order
against seeded values, so their precedence and the preset-folder guard cannot
regress. It also replays the wizard folder actions in sequence order (a
remembered folder beats the new-install default, a typed folder beats both) and
requires the machine restore action between the scope page and costing.
The author's new head 9bb0ecec97 is our fix-up 0b09eb43a5 plus main dacecbfea1, tree for tree, so this merge changes no file; it keeps #9455 marked merged with the batch.
CodeRabbit on #9775: the 12-of-181 count changed today, so ui-parity.md and gaps.md get the date in their status lines and a revision-history row, as AGENTS.md asks.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
CodeRabbit on #9775: the Linux portal swaps the extension for the picked type, but macOS and rfd keep the suggested name, so an opened image.pgm could still be saved as image.pgm holding P6 data. The suggested name itself now takes .pnm for any Netpbm subtype; the services test fails without it.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
CodeRabbit on #9775 (nitpick): #7689 enables flate2's runtime_detection feature, which 1.1.9 and older don't have, under a "1" requirement; a lock-less consumer (the publishable psd crate) could resolve an older 1.x and fail. Cargo.lock already pins 1.1.10 and is unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The pre-signing check matched any folder named Program Files, so it rejected
a valid MSI when Windows resolves Program Files elsewhere, accepted the 64-bit
root for an x86 package and accepted a different app name. Build the expected
folder from the package architecture's resolved root and the policy name.
Conflicts with main's batch #9880:
- psd file.rs / layer.rs: #5287 (PSB by encoded size, RLE tables widened on
promotion) against #6583's lazy reader. Main's new widening and length code
read `.data`, which #6583 leaves empty while the bytes stay in the source;
it now goes through `encoded()` / `encoded_len()` (also the auto-merged
lines in `rle_widening`, `encoded_len` and `LayerRecord::write`), and the
lazy reader sets `write_as_psb: false`. New test
`a_lazy_psd_promotes_to_the_same_psb_as_an_eager_one`: a lazily read PSD
writes the same PSB bytes as an eager one; it fails with a `.data` read.
- direct_select.rs / vector_ui.rs: #9321's Type Path and #6353's hidden
Work Path both kept.
- io tests main.rs: both modules. scorecard.md regenerated (12 of 181);
ui-parity.md and gaps.md follow it.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
`for_bands_grouped` always split the bands into `min(workers, jobs)` static
chunks, so on the default develop (jobs == workers) a worker held up on the
shared pool kept a fixed share of the bands waiting, where main's
`chunks_mut_limited` runs one rayon task per band. Now, when every worker may
hold a scratch, each band is its own task and borrows a scratch from a pool
(`par::chunks_mut_pooled`, at most one per worker, dropped as soon as no
waiting band can use it); the grouping stays for memory-limited jobs.
The fused bilinear/MHC, X-Trans and LinearRaw paths ran one band per worker
with no memory cap: a 262144 x 455 Bayer frame with bilinear needed 2.92 GB
at 15+ workers and failed with LimitExceeded under the default 2 GiB.
`working_memory` now caps those bands like AHD's and the fused paths use the
returned jobs. The fused LinearRaw and X-Trans band buffers are `band.min(h)`
rows, as charged (CodeRabbit, develop.rs:414).
Tests: a_slow_band_does_not_hold_up_the_bands_after_it and
fused_bands_at_once_are_capped_by_the_memory_limit fail without the fix;
for_bands_keeps_at_most_jobs_scratches_alive, chunks_mut_pooled_makes_one_state_per_worker_at_most,
multi-band and short X-Trans/LinearRaw fused-vs-full cases.
Co-authored-by: Valentin Klein <271991477+vangel76@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CodeRabbit on #9775: Place::AutoEditable puts type, shapes, pastes, placed files and frames above a group whose lock would refuse them; ui-parity.md's status line and checklist row said every new layer goes into an open active group.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
CodeRabbit on #9775: save_filters adds an unlisted suggested extension as its own type, so Save As on an opened .pgm offered .pgm again, though the encoder writes P6 for an RGB document. A suggested .pbm/.pgm/.ppm/.pam/.pfm now starts on the Netpbm type as .pnm; the services test checks every subtype name and fails without this.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
CodeRabbit on #9775: FontLibrary::finish wrote a failed type.font.favorite to the status bar as plain information; it now sets status_error like the other status writes. The regenerated scorecard counts 12 of 179 preferences that do nothing; ui-parity.md and gaps.md still said 178.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
CodeRabbit on #9775: the encoder picks P5/P6/P7/PFM from the image, not the name, so a .pbm, .pgm or .pfm picked in Save As held other data (an RGB document as P6 in a .pbm). Only the neutral .pnm is offered; File > Open still reads every Netpbm extension.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
CodeRabbit on #9775: newVideoLayerFromFile placed the layer while its content was still empty, so an active artboard always took it and clipped footage centred elsewhere; the first frame is now the content before placement, as in frames_to_layers (footage_off_the_active_artboard_stays_out_of_it fails without it). The merge also left a second check_group_depth ("Duplicate Layers") before main's "Copy Layers" one in layer.copyToDocument, so main's message could never show; it is removed.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
CodeRabbit on #9775: paint.stroke with fill:true builds the coverage, its feathered copy, the selection read and the mask over the region, up to the whole canvas (file.new allows 300000 x 300000), with infallible allocations that abort the process when memory runs out. The peak is now reserved through the fallible allocator first, and the selection is read through it; fill_survives_allocation_failure fails without this (the error came from a later blend).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Main's #9849 lets dropdowns grow back after a short list, and its test
`clearing_a_query_that_matches_nothing_grows_the_menu_back` expects a
search with no matches to shrink the font menu. #6526's menu stayed at its
full 534 px: its font list never shrank (auto_shrink y off), and the
vertical separator between the class list and the font list took the row's
whole available height. The list now shrinks, and the line is drawn over
the laid-out row instead, so the menu ends with the class list.
The test now checks that the menu ends with the class list (it can't drop
below half with the class list beside the list), and a new test pins that
the wheel past the end of the list, or over the class list, never moves
the header, also in a window too short for the full menu.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Main's #9849 bans egui::ComboBox; the preset menu #3558 added now uses
craft_ui::ComboBox, three characters past rustfmt's 160-column limit.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
merged_planes scanned each composite band twice on the calling thread
for alpha, encoded every range into per-channel Vecs one push at a time,
then copied them into the planes. The workers now write their slice of
every plane directly, check alpha as they go, and run a loop
monomorphised per sample type. The 8-bit deinterleave of layer pixels
used one thread per channel with a strided iterator; it now splits the
pixels into one band per core and writes each plane's slice in place.
Same arithmetic, same bytes; a test pins merged_planes against the
previous per-band implementation for RGB, Gray, Lab and CMYK at 8, 16
and 32 bits, with and without the matte.
bench_export .psd, 36 MP, 3 layers, 16 interleaved rounds of 7 reps
(median of medians / min), Ryzen 16 threads, Linux:
709.4 / 413.7 ms -> 540.8 / 318.9 ms (-24 % / -23 %)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A ZIP save encoded the thumbnail and the 1024 px composite preview as PNG
on the calling thread before any tile was compressed. The composite alone
is a ~200 ms single-threaded deflate on a 36 MP document, about 40 % of a
full save. The previews are now encoded on their own threads while the
document is collected and its new objects are compressed on all cores;
the bundle is assembled afterwards in the same order, so the bytes are
unchanged. Directory saves encode them up front as before. wasm stays
sequential.
bench_export, 36 MP, 3 layers, 16 interleaved rounds of 7 reps (median
of medians / min), Ryzen 16 threads, Linux:
.pcraft 517.9 / 471.1 ms -> 319.3 / 287.8 ms (-38 %)
.pcraft with cache 360.1 / 334.1 ms -> 277.5 / 268.3 ms (-23 %)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>