Repository navigation
docs: refresh the README "Tested against" benchmark - #85
Merged
Merged
Conversation
The figures were two deliverables stale and the benchmark walked staging. scripts/benchmark.py now walks only the curated disc states (active/archive/blocked), skipping incoming/ (staging and the E-mu .exb oracle corpus) and oracles/ — so it counts the six disc sources (106 images), not the 60 loose bank trees D35 converts, which have their own README section. It also gains an HFS (SampleCell) and a Kurzweil (KMSI) row in the by-filesystem table; both filesystems were converting but absent from it, so the table now sums exactly to the sample total. README "Tested against" is regenerated against the 106-disc collection: 102 of 106 converted, 150 640 samples, and — the headline change — the "entries not the file their entry placed" count falls 104 → 2 as D38's recovery lands (the paragraph is rewritten to match, naming the two that stay refused). Byte-identity 126 351 payloads, 170 305 WAVs written, rates across 1 761 values; the SampleCell discs' machfs check is noted. Closes #84. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #84. The "Tested against" figures were two deliverables stale, and refreshing them turned out to need a benchmark fix, not just a table paste (that's why it was deferred out of the 0.5.0 release).
scripts/benchmark.pyactive/archive/blocked, skippingincoming/(staging + the E-mu.exboracle corpus that D34's stereo verification uses) andoracles/. Pointed at the whole shelf before, it counted the 60 loose E-mu bank trees from D35 as "disc images" (166 total) and listed ~60.exbnames as pseudo-sources. Those loose banks are their own README section, not disc images from the six sources.KMSIrows — both filesystems were converting (D28/D29, D32/D33) but absent fromFS_ORDER, so their ~11.6k samples sat in the total with no row. The table now sums exactly to the sample total, and itsSkippedcolumn decomposes the headline damage count (AKAI 25 + Kurzweil 7 = 32 = damage 30 + not-placed 2).README "Tested against"
Regenerated against the 106-disc collection:
AMG - Loop Soup,Library.3'sCHINA2-R) named.machfsbyte check is noted.The EBL/AIFF prose and the short-image/displaced-partition narrative (D20) are unchanged — still accurate.
Verify
ruff check·ruff format --check·pytest -qall pass. The figures come fromuv run python scripts/benchmark.py ~/…/sample-images(a real batch run over the collection).🤖 Generated with Claude Code