Repository navigation
Repo hygiene: history & tip cleanup notes (low priority, measured 2026-10-31) #3204
Description
Low priority — documentation only, no action scheduled. Measured 2026-10-31 as part of a fleet-wide repo hygiene census. Full source data: repo-history-census-2026-10-31.md and repo-inventory-2026-10-31.md (sandbox workspace, ~/workspace/).
Repo snapshot
- Commits: 5,532 on
main— first commit 2023-04-17, last 2026-10-31 - Visibility: public
- Fresh clone of
main: ~78.4 MB pack (GitHub repo size 79 MB) — tip is 5,510 files / 67 MB - Local clone: 217 MB
.git - Open PRs: 53 (re-measured 2026-10-31)
Junk in history (main-branch history, measured by blob census)
337 MB uncompressed / ~26 MB packed, almost all in 65 objects — stripping them would shrink a fresh clone from 78.4 MB to ~52.4 MB (saves 26.0 MB, 33%). Nearly all of it is already dead history (335 of 337 MB is not at the tip):
- Roc build outputs, recommitted across versions:
pkgs/roc/**/build— 133 MB + 91 MB across versions;platform/libhost.a— 66 MB total (11.7 MB × 5 versions);dynhost,*.rh - Compiled binaries, committed then deleted:
pkgs/launcher/launcher(17.4 + 15.6 MB),pkgs/database/database(12.0 MB),pkgs/import/import(8.7 MB),pkgs/chat/chat(7.9 MB),pkgs/tea/tea(7.6 MB) - One-offs:
infra/dns/terraform.log— 20.9 MB (the single largest blob in the repo's history is a log file);sources/map/data/data.kz— 6.5 MB - All-refs census (including side branches): 837 MB of the 900 MB total blob bytes are not at the tip.
At the tip today there is exactly one file over 2 MB: doc/NanoPi… at 12.5 MB.
Verdict: NO history rewrite
26 MB packed does not justify orphaning 53 open PRs, invalidating ~150 fleet worktrees, and changing every commit SHA cited across the fleet's issues and docs. As a public repo, a rewrite also wouldn't erase the old history — forks, existing clones, and GitHub's PR refs (refs/pull/*) keep the objects alive until GitHub's own GC.
Already done
- Nothing needed at the tip: it is already clean except for the one
doc/NanoPifile below. The junk is history-only.
Open questions (human calls, no deadline)
- Roc platform artifacts — intentional or accidental?
pkgs/roc/**/platform/libhost.a,*.rh,dynhost, and thebuildoutputs may be intentional platform distribution artifacts (Roc platforms ship prebuilt hosts) rather than accidental commits. If accidental: untrack + gitignore the classes at the tip. If intentional: document that here and close the question permanently. doc/NanoPi(12.5 MB at tip) — keep as a doc asset, or move to release/artifact storage?
Standing forward-hygiene rules
- No build outputs, compiled binaries, or log files committed at the tip; build artifacts belong in releases or artifact storage (subject to the Roc question above).
- Gitignore scratch classes (scratch/output dirs, API/monitor dumps, probe dirs,
*.err/*.out). - Runtime state files belong outside repos (
~/.local/state/) per the pm-aggregator precedent; repos hold summaries. - Re-run the hygiene census quarterly, or sooner if this repo's main pack crosses ~250 MB or junk classes appear at the tip. Fleet tracking issue: "Standard task: per-repo hygiene census + cleanup documentation" (this repo).
Acceptance criteria
- The measured census for this repo (blob census, junk classes, fresh-clone size, verdict) is on record in this issue.
- No history rewrite is performed: the recorded verdict stands unless a future census shows a materially larger saving.
- Forward hygiene: no build outputs, binaries, or log files are committed at the tip going forward.
Activity
dezren39 commentedon Oct 31, 2026
Intake review
Fleet monitor intake pass (protocol §5), 2026-10-31.
Related: Per-repo hygiene notes for developing-today/code; fleet tracking issue: code#3205. Sibling notes: opencode#1101, openchamber#1069, nix#1265.
Notes / risks / suggested approach: Documentation only, no action scheduled. The one tip-level item noted is
doc/NanoPi…at 12.5 MB — the only file over 2 MB at the tip. Risk: none from leaving as-is; a rewrite would orphan 53 open PRs to save ~26 MB packed. Suggested approach: none — keep on record; re-measure at the quarterly census.dezren39 commentedon Nov 3, 2026
agent-note:v1 — task 8215 handed off (Lane A worker, 2026-11-03): execution lane assigned opencode; the claimant launches the bound session to finish this task. Prior progress on the task is at stage triage-verified (verified at main 5d38fdc18): findings — one tracked file >2MB (doc/NanoPi, 12.5MB), no tracked build outputs/binaries at tip, 5970 tracked files, tracked .log files only under a doc spikes folder as intentional evidence; outcome recorded: no implementation path, the issue itself is the deliverable and its acceptance criteria are met. The bound session posts the full triage note and completes task 8215 as triaged.
dezren39 commentedon Nov 3, 2026
🤖 Agent note · handed off
ses_e5e5dc4471dffeKS8jvrsYNigYrepo-worker-code(bound worker uses this id for all queue helper verbs)opencode-route/workhorse, agentbuilddezren39 commentedon Nov 3, 2026
Triage verdict: refreshed measurements, no cleanup action
Rechecked current
mainat1c1aa4a799fc8e11ff208118230de955996e509c(2026-11-03, queue task 8215, workerrepo-worker-code). The 2026-10-31 figures have moved with normal forward churn; the standing decision is unchanged: no history rewrite, and no tip cleanup needed.Current read-only measurements
main; first commit still 2023-04-17 (was 5,532 — growth from merged branches, not a rewrite).Tip-level junk check at
origin/maindoc/NanoPiat 13,078,807 bytes (12.5 MiB) — the known open question Bump checkmarx/ast-github-action from 2.0.11 to 2.0.18 #2, unchanged.pkgs/cold-store-reaper/worker/dist/worker.js(20,612 bytes). This is intentional and documented — it is the deployable Cloudflare Worker bundle (pkgs/cold-store-reaper/worker/README.md: "dist/worker.js — this is what gets deployed", "the operator deploys dist/worker.js as-is"). Not junk..log/.err/.outfiles totalling 2.33 MB, all under thedoc/2026-10-11T…spike and research-evidence folders. Small aggregate records, not the 20.9 MBterraform.logclass the census flags.Triage outcome
No implementation path. The issue is itself the deliverable (census + no-history-rewrite verdict) and its acceptance criteria are met:
The two open questions (Roc platform artifacts;
doc/NanoPi) remain human calls with no deadline and are recorded in the issue body. The quarterly re-census stays with the fleet tracking issue #3205. Closing this per-repo record as completed, consistent with siblingsdeveloping-today/opencode#1101anddeveloping-today/openchamber#1069.Blockers / ordering: none. No repo state is changed by this triage and no follow-up queue task is required.
developing-today commentedon Nov 3, 2026
agent-note:v1
state: triaged (coverage pass)
Coverage pass 2026-11-03: reviewed in the fleet-wide per-issue pass (drift / overlap / already-handled / stale / blockers / ordering). Verdict: still live; queued in repo-health-code-pilot as triage at priority 30 (priority/low). If this is already handled at current main, the triage closes it with evidence; if it needs splitting, the triage files the split.
Found via: coverage pass 2026-11-03