Repository navigation
Release: v0.10.2 - #254
Merged
Merged
Release: v0.10.2#254
Conversation
gitsheets#301 shipped repo.sync() and repo.isAncestor in 2.8.0 and develop already runs ^2.9.0, so the external block is gone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
…anism Mechanism-only wording: reconcile moves from hand-rolled git plumbing to gitsheets' sync primitive, and the hot-reload pre-check to repo.isAncestor. Outcomes, conflict-branch naming and the single write-lock critical section for ref move + rebuild are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
Swap reconcile.ts's hand-rolled merge-base / merge-tree / commit-tree /
update-ref / push plumbing for repo.sync({ onConflict:
'preserve-branch' }), and the webhook's rev-parse + merge-base
pre-check for repo.isAncestor. Outcome vocabulary and webhook response
are unchanged: gitsheets' 'ahead' maps to 'pushed-ahead' and the
conflict branch keeps the conflicts/<UTC> name.
repo.sync() takes its handle's write lock and throws lock_held inside
withLock, but the hot reload needs ref move + store refresh + rebuild
in one critical section, or a write queued on the lock can run in the
gap, validate against pre-reload in-memory state and clobber what the
reload brought in. So the serving repository's withLock still wraps
the whole reconcile, and sync runs through a short-lived side handle
on the same gitdir whose own lock is uncontended. The fetch stays a
separate lock-free step (sync can't fetch without locking).
A new test holds the lock, queues the reload and then a write behind
it, and asserts the ref doesn't move until the lock frees and no
commit lands before the rebuild; it fails for both sync-then-lock
variants.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
A staff spam vote only deactivated; deleting took a pipeline run or a Purge click per account. A daily in-process sweep now purges everyone whose human votes are all spam, the newest three or more days old, in one commit, and the roster shows when each will go. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TGqwPjt3v4eMfiHpPmCKnR
Members whose human votes are all spam, still deactivated and not
staff, are purged by a daily in-process sweep three days after the
newest vote: all of them in one cfp-api commit listing the slugs, with
the projects only they touched. The purge cascade moved into
purgePeople() so the Purge button and the sweep share it. Roster rows
carry purgeAfter, shown as a "Purges {date}" chip.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TGqwPjt3v4eMfiHpPmCKnR
Plan the follow-up to gitsheets-scripts now that gitsheets 2.9 ships ref-pinned sheets (#303) and public raw-file writes plus orphan and branch-creating transactions (#305). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
… transact
gitsheets 2.9 ships ref-pinned read-only sheets (#303) and lets
transact({ parent, branch }) create a missing branch (#305). Read
person-evaluations through openSheet(name, { ref }) instead of a no-op
transaction, and parent a fresh clone's first prune on origin/<branch>
so gitsheets creates the local branch, dropping the create-only
update-ref.
A dry-run or no-op run on a fresh clone no longer creates the local
branch; it appears only when a prune commits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
…transact
Read the previous snapshot's ids through openStore(repo, { ref }) (#303)
rather than a no-op transaction, and on first run parent the snapshot on
origin/<branch> or the initial parent so transact({ parent, branch })
creates legacy-import (#305). The create-only update-ref is gone.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
Replace writeConfigRootCommit's hash-object / temp index / write-tree /
commit-tree plumbing with one orphan transact({ parent: null }) that
stages the source's .gitsheets/ files with tx.writeFile (gitsheets
2.9, #305).
The final snapshot commit stays on raw commit-tree + update-ref:
2.9's tx.sheet() loads configs from the transaction's parent (HEAD for
an orphan), never from files staged with tx.writeFile in the same
transaction. So records can't be upserted straight into an orphan, and
gitsheets can't point a branch at an existing commit. Each remaining
raw call is commented with its reason.
Tests now pin the snapshot shape: a parentless commit carrying the
source's configs and contracts byte-for-byte plus records, HEAD on the
snapshot branch, the tag, a no-op same-seed re-run, and no refs after
a failed verification.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
Records that scrub-data's single-transaction snapshot isn't reachable on gitsheets 2.9 (tx.sheet() can't see configs staged with tx.writeFile), lists every remaining raw git call in apps/api/scripts with its reason, and files the upstream gap plus the two backfill scripts as follow-ups. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
The intro still said "the next pipeline run removes them"; a spam vote now leads to a purge three days later unless someone disagrees. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TGqwPjt3v4eMfiHpPmCKnR
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TGqwPjt3v4eMfiHpPmCKnR
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011Q2tgjZEUrtvVk6k8SpdXJ
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TGqwPjt3v4eMfiHpPmCKnR
refactor(api): reconcile through gitsheets repo.sync()
Author
Changelog- refactor(api): reconcile through gitsheets repo.sync() [#251] @themightychris
- refactor(scripts): read pinned sheets and create branches through gitsheets [#252] @themightychris
- feat: daily scheduled purge of spam-voted accounts [#253] @themightychris |
refactor(scripts): read pinned sheets and create branches through gitsheets
feat: daily scheduled purge of spam-voted accounts
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.
Technical