Conversation
Add scripts/demo/ to record the oans demo GIF in one command, deterministically: - setup.sh — builds a demo tree on btrfs/xfs: mostly unique files (hashed, long scan phase) plus a small set of duplicated groups (short dedupe phase), with exponential file sizes (many small, few large) from a fixed seed. Copies use --reflink=never so oans actually has something to reclaim. Defaults: ~3000 files, ~6 GiB, 100 dedupe groups. All knobs are env-overridable. - record.sh — one entrypoint: builds oans, generates the dataset in the current directory (short paths in --stats/--history), records with VHS if installed else asciinema+agg, then gifsicle-optimizes the GIF. Cleans up the dataset. - demo.tape — the VHS screenplay (theme, font, timing; waits for the summary). - play.sh — the same screenplay for the asciinema+agg fallback. Not wired into the README yet — that follows once a render is committed. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Generating tens of thousands of files with one head/cp per file was slow. Now: - one awk pass plans every path + exponential size (no per-file shell subshells), - a pool of python workers (one per core) each open /dev/urandom once and stream bytes for many files — no per-file process spawns, - copies are made with a single parallel `cp` via xargs -P. Measured ~0.26s for 3150 files (32 cores) vs seconds of spawns before. Also drop the page cache after generation (best-effort sudo; DROP_CACHES=0 to skip) so the recorded scan reads cold from disk — realistic, and it makes the hashing phase visibly longer. Verified: correct file counts and no accidental dedupe of the unique files (oans dedupes exactly DUP_GROUPS groups). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Replace the inline python heredoc in setup.sh with a real, lintable scripts/demo/gen.py, and fill files with Python's random (Mersenne Twister) instead of /dev/urandom — plenty for a demo (bytes only need to be distinct and incompressible). Each parallel worker seeds its RNG from its chunk filename, so the streams are independent and two "unique" files can't collide (verified: oans still dedupes exactly DUP_GROUPS groups). ~0.32s for 3150 files. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The tape waited for /Reclaimed/, but the dedupe summary's headline wording varies by oans version (e.g. "Deduplicated ... / Kernel scanned ..."), so on some builds the match never fired and VHS hung until the 300s timeout. Wait for "Elapsed" instead — printed on every dedupe run regardless of wording — and bump the cap to 600s for very large trees. Verified end-to-end: render completes, no timeout. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Terminal casts read fine at 12fps, and halving the framerate roughly halves the frame count — the main GIF-size driver, since the scan progress bar changes every frame. Also expose the gifsicle --lossy level as DEMO_LOSSY (default 60, was 30). Measured ~25% smaller on a warm run; more on a long cold scan where the changing frames dominate. Text stays crisp. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Commit the generated demo GIF (assets/demo.gif) and embed it as the hero image near the top of the README, sized to 900px so it maps 1:1 to GitHub's content column. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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
1 participant
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.
Adds
scripts/demo/— a one-command, reproducible way to record the oans demo GIF for the README. No manual typing, fully re-runnable.Usage
What's here
setup.sh— builds the demo tree on btrfs/xfs. Deliberately shaped so the scan/hashing phase is the long, visible one and the dedupe phase is short: ~2700 unique files (hashed, never deduped) + 100 small duplicated groups. File sizes follow an exponential distribution (many small, a few large) from a fixed seed. Copies use--reflink=never(on btrfscpreflink-copies by default, which would leave nothing to reclaim). Defaults ≈ 3000 files / ~6 GiB / 100 dedupe groups; every knob is env-overridable (DEMO_UNIQUE,DEMO_DUP_GROUPS,DEMO_COPIES,DEMO_MEAN_KB, …).record.sh— builds oans, generates the dataset in the current directory (so--stats/--historyshow short paths), records with VHS if installed (highest quality) else asciinema + agg, then gifsicle-optimizes the GIF (~30% smaller, quality-neutral). Cleans up the dataset afterward, exposes oans viaPATH(never copies/deletes your binary).demo.tape— the VHS screenplay (Dracula theme, Hack font, 1280×760;Wait+Screen /Reclaimed/so timing is correct regardless of dataset size).play.sh— the same screenplay for the asciinema+agg fallback.Notes
assets/demo.gifis committed.🤖 Generated with Claude Code