This is the actual bug the night's ecosystem-utils audit was looking for.
classifier.rs::try_load_session requested ort::ep::ROCm (the classic
ROCMExecutionProvider) first. Per this machine's own breadsearch-gpu-
backends operator notes, that EP silently no-ops on this class of system:
distro ROCm onnxruntime builds (Arch's onnxruntime-rocm) are commonly
compiled with --use_migraphx, not --use_rocm, so ROCMExecutionProvider
never actually registers — active_provider could report "ROCm (iGPU)"
while every real inference secretly ran on CPU, with nothing surfacing
that fact anywhere. Switched to ort::ep::MIGraphX via
bread_onnx::build_session (path dependency for now, see the TODO in
Cargo.toml), matching breadsearch's own already-correct embed.rs.
Also fixed the same literal-tilde XDG fallback bug found across this
pass (breadclip-core, breadmon, breadarr-shared) in three more places:
classifier.rs::model_dir, config.rs::config_path, config.rs::style_css_path.
Bumped the workspace's tokenizers pin 0.21 -> 0.23 to unify with
bread-onnx's own requirement (breadarr already pins 0.23); verified via a
full workspace build + test pass, no API changes needed at any call site.
Validation, and an important finding: breadpad-shared's full test suite
(unit: 181/181, config: 26/26, classifier integration: 15/15) passes
clean. The pipeline.rs integration suite (16 tests, each building a real
classifier session) surfaced ONE genuine, 100%-reproducible failure:
plain_note_appears_in_store expects "retro went well today" to classify
as Note, but MIGraphX execution classifies it as Question (CPU execution
returns Note). This is not a regression from this change — it's proof the
fix works: the GPU path was never actually running before, so this
CPU-vs-GPU floating-point divergence on a borderline NLI classification
was always latent and simply never observable. Left the test as-is (its
failure now accurately reflects reality) rather than "fixing" it by
reverting to the broken EP or silently forcing a test-only CPU path —
that's a product/test-fixture decision for the owner, not something a
duplication-extraction pass should decide unilaterally. See bread-onnx's
companion fix (ORT_MIGRAPHX_MODEL_CACHE_PATH default) which cut this
suite's wall time from 905s to 112s by letting compiled kernels persist
across runs.
Unifies the "selected" visual language across breadman's sidebar and
breadpad's type chips (ghost default, solid accent when active), gives
note cards a real surface instead of just a left border, caps the note
list width so type chips and action buttons sit next to the title they
belong to, and swaps the sidebar's full-color emoji for theme-safe
monochrome glyphs plus per-type colored dots.
breadman also gets live per-row sidebar counts, a note-count footer,
relative timestamps with absolute-time tooltips, and a de-emphasized
Errors row.
breadpad's popup gets an explicit short-prefix grammar (td:, rem:,
idea:, note:, q:) that live-highlights the matching chip as you type,
a keyboard-shortcut hint row, and a confirm button that's visually
separated from the chip row instead of reading as a sixth pill.
Switches ort from load-dynamic to rocm feature so the ROCm execution
provider is compiled in. Adds rocm-runtime to bakery system_deps so
bakery doctor/install can verify it's present.
Eliminates the spurious ERROR log from ORT when ROCm isn't compiled in.
Checks is_available() before attempting registration so the session
correctly falls back to CPU without noise.
- Add MIT LICENSE file
- Expand .gitignore with standard Rust/Linux entries
- Remove dangling symlinks (breadmancli, breadpadcli) and dev scratchpad (svgs.txt) from git tracking
- Replace unsafe unwrap() calls with expect() in breadman CLI (guarded by prior filter)
