The probe asked for a file's size, read it (an open and its own size question), and then asked for its modification time again by path: three round trips to a network share's server for what one open can answer. It now opens the file once and takes the size and the modification time from the open file. What is read, the work-gate weight, and the text of the error for a file that can't be read are unchanged.
- RawImage::develop demosaics only the default crop with AHD when there
is no OpcodeList3 (ahd_window), instead of demosaicing the margin too
and then copying the crop out of the full picture on one thread.
- Gpu::upload writes through queue.write_buffer_with and copies into
the staging buffer in parallel parts. create_buffer_init zeroed a
staging buffer and then copied over it, two single-threaded passes
over a full-size photo's ~290 MB.
- finish_and_read copies the mapped readback out in parallel parts (one
thread read the 91 MB of a 24 MP export at ~2 GB/s, ~45 ms).
Output is byte-identical (ahd_window_is_the_crop_of_the_whole,
uploads_read_back_exactly).
Export of a 24 MP ARW to full-size JPEG (examples/export_bench, 75
interleaved runs each on a shared 16-thread Ryzen):
main 802 ms median / 614 ms min -> this branch 647 / 525 (-19.3% / -14.5%).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- finish_8bit/finish_deep hand back the render itself when there is no
output sharpening and no watermark. Before, every export cloned the
image (~100 MB at 24 MP) and output_sharpen built a ~290 MB float copy
of it even for SharpenFor::None, all on one thread.
- Output sharpening's box blurs run row-parallel (par_rows); each pixel
is computed as before.
- The parallel JPEG encoder converts each band's rows to YCbCr inside the
band's task instead of first filling whole-image planes (~290 MB for
4:4:4 at 24 MP, written once and read back).
- fdct_quant rounds with plain arithmetic (round_i16) instead of
f32::round, a libm call per coefficient without SSE4.1; it equals
`x.round() as i16` for all 2^32 inputs.
Output is byte-identical (pinned by encoded_bytes_are_pinned and
sharpened_pixels_are_pinned, whose hashes come from main).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
examples/export_bench imports the files given and runs export_photo
(decode, full-size render, encode) N times, forgetting the decoded
original between runs like a batch export does, and prints the median
and minimum wall time. OPT takes export options as app.export JSON,
PHASES=1 times render and encode separately, OUT=dir writes the files
for byte comparisons between builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Use craft-ui's dropdown so menus grow back after shrinking
The album picker keeps its choices and search field in a fixed header and scrolls only
the list; combo_box and the print dialog's combos use craft_ui::ComboBox. egui::ComboBox
is now disallowed by clippy.
* chore(deps): craft-libs 60d60d48 (dropdown fixes)
* chore(deps): craft-libs 6af63481 (dropdown fixes)
* chore(deps): craft-libs 4d7b82f7 (dropdown fixes)
* chore(deps): craft-libs 92d392cf (#387 merged)
Reuse cached folder listings through Arc and allow tests to inject Local places.
Show browsed folders below hidden ancestors as session roots. Apply the same
hidden-entry checks to folder walks, browsing, import scanning, and Local.
Return an error when either file index is zero. Keep the resolved-path fallback for lenient comparisons. Test both unknown-identity orderings and distinguish identity lookup failures from known-original refusals.
Merge main at 662df66f.
Nathaniel Ledford