The two clones had diverged from origin/main (191d1b8) and both had
grown uncommitted work. This merges the other clone's three commits and
reconciles four conflicting files.
Both clones independently grew a feature called "batch export", but they
are different sweeps and both are kept:
- BatchExportWorker (this clone) — every angle x channel of the one open
file, fixed colorbar per channel, under a new Export menu.
- BatchExportImagesWorker (other clone) — the current view across many
selected files, process-pooled, under Convert.
Conflict resolutions of note:
- sras_average.py: the other clone's memory-bounded rewrite had silently
reverted this clone's float32 -> int16 accumulator fix, which exists to
keep averaged output from landing 1 ADC count off the original tool.
Kept the rewrite, re-applied the fix, and corrected the two docstrings
that still claimed float32.
- tests/test_compute.py: kept the other clone's meta["waveforms"] key
(meta["data"] no longer exists for this fixture and would KeyError)
and its laser_freq_hz assertions, plus this clone's int16 expectation
and its dropped subprocess returncode assertions.
- ComputeWorker: row_avg_n and min_freq_mhz were added independently on
either side; both are now threaded through.
- compute_rf_image's `exact` parameter is gone, removed by the other
clone's PyFFTW peak-search rewrite. It had no remaining callers.
Verified: 149 passed with a complete dependency set. The failures seen
with this clone's own .venv are a missing scikit-image, which predates
this merge and reproduces identically on the pre-merge commit.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
origin/main split sras_viewer.py into the sras_viewer/ package (dialogs,
main_window, common, canvases) after the batch-export feature was built
against the old monolith, so merging origin/main into main lost the feature
to the sras_viewer.py deletion. Re-add it in the new structure:
BatchExportDialog in dialogs.py, a Jobs.EXPORT job key in common.py, and
the menu action/worker wiring in main_window.py, following the Jobs-key
and dialog-class conventions the split already established rather than the
old ad hoc "main" progress-key naming the original commit used.
Also carries forward two test fixes bundled in the original batch-export
commit that the parallel pytest conversion (007089d) had ported from
tools/test_refactor.py without them: the int16-vs-float32 accumulator
expectation in test_sras_average, and asserting the sras_average.py
subprocess calls actually succeed.
The old tool only understood the legacy uniform-geometry header and had no
bound on peak RAM, making it unusable on the ragged v6/v7 files the scanner
now produces at up to ~560GB. Reads via SrasFile's memmap and writes in
row-chunks sized to a memory budget (SRAS_MEM_BUDGET_MB), stages to .part and
os.replace()s per scan_format.md's atomic-write requirement, always writes
plain v6 (a v7 cache tail is frame-indexed and invalid after averaging), and
divides laser_freq_hz by N so x_axis_mm() stays correct after the X axis gets
spatially binned.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Drop the coarse+fine zoom refinement, the SciPy FFT backend, and the
exact= audit path in favor of a single always-on full-transform peak
search (_peak_bins). Block size is now derived from a per-thread memory
budget (_fft_block_for/SRAS_FFT_PLAN_BUDGET_MB) instead of a fixed
constant, so the existing block-parallel PyFFTW pool stays memory-safe
at high pad factors without the zoom algorithm's bookkeeping. Also
removes the now-unused threadpoolctl dependency and the FFT backend
selector from the UI.
Also includes a pre-existing min_freq_mhz peak-search floor (excludes
bins below a caller-supplied frequency from the argmax) that was
already implemented and tested in the working tree.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- FFT backend and pad factor persist across sessions via QSettings
(IniFormat; tests redirect the settings path for hermeticity).
- The default backend was labelled "NumPy FFT" but always dispatched to
scipy.fft — rename the canonical value to "scipy" ("numpy" stays as a
legacy alias) and fix the dialog label.
- cache_file: DC caching fans out over angles via _parallel_map with
per-angle budgets (the DcPrecomputeWorker pattern); FFT caching stays
serial per angle because compute_rf_image now parallelises internally
over blocks. Documented that the v7 FFT cache is natural-resolution
(pad 1) by design.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
At pad 40 the old path materialised a ~9 GB padded spectrum per row,
which collapsed the chunk planner to one worker and one rfft call with
workers=1 — synthesis ran single-threaded, ~1 hour per angle on real
files.
The padded spectrum is never materialised now. Each block of 512
waveforms gets a coarse rfft at next_fast_len(2*spf); every coarse bin
within 0.7 of its row's max (plus the DC-adjacent window, which coarse
DC suppression would otherwise blind) is refined onto the exact n_fft
grid by a small complex gemm. The selected bin is bit-identical to the
full padded argmax — enforced by test_zoom_identity, a 25-seed fuzz
test over adversarial spectra, and a clean golden-hash diff against the
pre-rewrite baseline across pads {1,2,4,8,40}, masked/unmasked, bg
on/off, int8/int16, and both backends.
Blocks fan out over a persistent thread pool; pyFFTW runs through
per-thread FFTW_MEASURE builder plans with wisdom persisted to
~/.cache/sras-viewer, and threadpoolctl clamps BLAS under the pool.
compute_rf_image(exact=True) (or SRAS_FFT_EXACT=1) keeps the reference
padded path for audits.
tools/bench_fft.py measures: pad 40, 16 cores, 8192x2500 synthetic —
exact serial 717 wf/s -> zoom pool 25100 wf/s (35x, pyFFTW backend;
19x scipy), every variant verified equal to the reference.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- pyproject.toml replaces sras_viewer_requirements.txt (same pins) and
adds a dev extra with pytest.
- tools/test_refactor.py, test_alignment.py, test_gui.py become
tests/test_compute.py, tests/test_alignment.py, tests/test_gui.py with
assertions preserved verbatim. test_gui.py stays one ordered
integration sequence over a shared module-scoped window.
- check_equivalence.py: drop the dead pre-refactor monolith shim (and the
_compute_angle_alignment alias it consumed), extend the pad sweep to
(1, 2, 4, 8, 40), add legacy-v4 and big-endian int16 legs (new bps=2
option in make_test_sras) so the padded FFT path and the >i2 memmap
path are in the baseline before the FFT rewrite.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Thomas Ales [M S E]