grit://cooking/changelog← what’s cooking
Saturday, October 31, 2026 · release v0.5.27

92 changes landed on main

58 issue fixes · 34 other changes

v0.5.27 fixes many rebase, merge, reset and ref-handling bugs and speeds up large-repo operations such as reset --hard, ref listing and since-bounded log/revs.

  • ›Rebase continue/skip no longer drops the rebase after a conflict; merge-history rebase no longer duplicates commits (#2349, #2357, #2363, #2365)
  • ›Faster large-repo ops: packed-ref streaming, reset --hard, switch, and log A..B -n with a path or --since
  • ›Fixed merge bugs: directory/file conflicts, adjacent edits, final-newline loss, and stack overflow with sibling dirs (#2359, #2369, #2381, #2392)
  • ›Fixed GIT_INDEX_FILE writes, :(top) and :(attr) pathspecs, stash list --json, and unstaged diff rename pairing
  • ›Smart HTTP and SSH fixes: streamed large fetch bodies, protocol v2 over SSH, clone URL credentials, and --deepen
  • ›Added grit blame -M, typed library errors, and persistable workspace snapshots; nested repos now warn and fetch cleans up on TERM
$ git log f91a862b..v0.5.27202 commits on GitHub ↗300+ files +13,041 −2,983
Highlightsmeasured by the reviewers, before → after
grit log A..B -n 5 -- <path> costs 1.6-8 s on linux regardless of N and range size (git: 0.09-0.14 s)
50,021-commit synthetic range+path warm wall time (debug build; system Git reference 2 ms)169× better████████████████████████ 1,518 ms median of 5 █ 9 ms median of 5
grit merge/pick/rebase abort with stack overflow (infinite recursion) when conflicted dir `a/` has sibling dir `a0/`
merge process exit status (SIGABRT shell code vs expected conflict code)134× better████████████████████████ 134 exit code █ 1 exit code
upload-pack / branch listing on a files-backend repo with 1M packed refs: 2.0 s, 336 MB (git 0.16 s, 66 MB)
grit branch -n 1 median wall time on 1M packed refs82× better████████████████████████ 1.4 seconds █ 0.017 seconds
grit revs --since DATE walks the whole history: linux 31-70 s vs git rev-list 0.09 s (log --since is fine)
100k-commit synthetic history, 100 selected: origin/main grit vs review branch median72× better████████████████████████ 180.2 ms █ 2.5 ms
grit log --since combined with --author/--grep/pathspec walks all history: 1.3 s vs git 80-96 ms on git/git
git/git --since + --grep=fix + Documentation median (7 runs)20× better████████████████████████ 1,005 ms █ 49.9 ms
issues fixed58
  • ›grit stash list --json emits a bare JSON array; every other list command returns an object (v0.5.26)#234622dd4f2d

    `grit stash list --json` now returns an extensible object containing `action: "list"` and an `entries` array, including for an empty stash. Human and Markdown rendering remain unchanged, while the machine contract advances to a documented v2 schema and keeps v1 frozen.

    Empty stash-list median latency (debug build, 100 runs): 4.4 ms → 4.3 ms (1% better)Grit/system-git empty stash-list latency ratio: 4.8 x → 4.8 x (1% better)
  • ›grit diff (no args / --unstaged) pairs deleted tracked file with an untracked file as a rename, but omits untracked adds#2342f8543ad2

    Unstaged diff no longer scans untracked files or feeds them into rename detection. A plain filesystem move now appears as deletion of the tracked source, while the destination and unrelated new files remain exclusively in status as untracked paths.

    Grit unstaged name-status median, debug binary, 10,000 untracked files: 92.3 ms → 4.7 ms (20× better)
  • ›status inside an untracked dir prints 'newdir/' instead of './' (repo-relative); '..' dirs lose trailing slash#2344a6ee9dd7

    Human-facing status paths now remain meaningful when invoked inside untracked directories: the current collapsed directory is shown as `./`, ancestor directories as `../`, and all collapsed directory entries retain `/`. JSON remains unchanged and repo-relative.

    Release `status -s` median on 250 untracked directories: 3.0 ms over 35 interleaved runs (system Git: 1.815 ms) → 3.0 ms over 35 interleaved runs (system Git: 1.815 ms) (0% better)
  • ›grit diff --patch-text prints placeholder new blob id (0000000) in `index` line for worktree diffs#2341b704da15

    Patch rendering now derives blob ids from the actual resolved text or symlink bytes before writing `index` headers. Raw worktree diff entries remain unchanged, while binary behavior, absent sides, and gitlink commit ids retain their existing semantics.

    500-file optimized patch render median: 29.0 ms (12 runs) → 33.6 ms (12 runs) (14% worse)
  • ›clone/fetch write no reflog for remote-tracking refs: origin/main@{1} fails, git writes one#234334a2249e

    Fetch and pull now preserve prior remote-tracking tips in reflogs after branch updates, while initial clone still imports branch refs in one packed batch and logs only the remote HEAD. Pruning removes the corresponding reflog, and logging follows core.logAllRefUpdates/existing-log policy through the selected ref-store backend.

    Dist 101-ref local fetch median: 11.0 ms → 14.0 ms (22% worse)Dist fetch time relative to system Git: 0.822 x Git median → 1.0 x Git median (20% worse)
  • ›grit repack/gc leaves a second full 2 MB copy of an edited blob: pack 2x larger than git gc (v0.5.26)#23453c14b849

    This change prevents delta search from cutting a run of versions for the same path at an arbitrary byte ceiling. Edited multi-megabyte blobs therefore remain visible to one sliding window and collapse to one full base plus deltas instead of retaining a second full copy.

    repacked bytes: 4,276,920 bytes → 2,183,223 bytes (96% better)undeltified 2 MiB blob versions: 2 objects → 1 objects (2.0× better)release repack wall time: 0.199 seconds → 0.200 seconds (0% worse)pack-size ratio versus system Git: 2.0 ratio → 1.000 ratio (96% better)
  • ›grit never requests protocol v2 over ssh (no GIT_PROTOCOL env), ignores protocol.version=2; always uses v0 advertisement#2338e46dc720

    SSH upload-pack connections now request the configured Git wire protocol and correctly tell OpenSSH to forward GIT_PROTOCOL. Remote fetch/listing defaults to protocol v2 while configured v0 and receive-pack push remain on v0.

  • ›Ref advertisement for 1M refs costs ~420-540 MB per request and 3x git's time (serve + ssh upload-pack), no streaming#2348fced6cf3

    Ref advertisements are emitted incrementally instead of being retained as owned names and duplicate line buffers. Reftable stack files stay in shared file-backed mappings, and HTTP discovery now carries advertisement chunks through a bounded channel, limiting per-request anonymous memory.

  • ›grit serve: fetch/clone of a repo with more than ~650k refs fails with HTTP 413 (32 MiB fetch request cap)#234778b24ef3

    Smart HTTP fetch request bodies are streamed into upload-pack instead of buffered behind the former 32 MiB cap. Decoded fetch input is bounded at 1 GiB, including highly compressible gzip requests, and protocol v2 now preserves that transport limit as HTTP 413 rather than converting it to an HTTP 200 ERR pkt-line.

  • ›rebase --continue/--skip after conflict on the first of several commits says 'Already up to date', drops the rebase#23504a45220b

    Rebase resume planning now honors pending replay work and seeded mappings even when the edit map is empty. This prevents continue or skip after an early conflict from discarding the session as an empty rewrite.

  • ›rebase --continue with 2+ commits left to replay says 'Already up to date', undoes the rebase, loses the resolution#234967732b6c

    Rebase continuation and skip now retain partial replay state when the filtered edit map is empty but seeded mappings or remaining commits still require work. Human and Markdown output also reserve “Already up to date” for true no-op outcomes with no head movement.

  • ›grit commit during a rebase conflict, then rebase --continue fails 'nothing to commit after conflict resolution'#229398cb3a9d

    Resumed rebase planning no longer mistakes an empty explicit edit map for an empty operation. Continue and skip now replay the remaining commits and finish the branch when the first commit in a multi-commit rebase was the conflict stop.

  • ›grit-lib switch::resolve_create_branch returns Error::Message strings for already-exists / invalid-name / no-commits#23528892acb2

    Switch branch creation now exposes existing typed library errors for duplicate names, invalid names, and unborn repositories instead of forcing embedders to parse English text. Revision start-point failures are also typed consistently, and switch JSON output classifies the new branch errors with stable kinds.

  • ›grit log/revs --reverse -n N shows the wrong window: drops the newest commit, adds an older one (also --skip, --json)#2354375f5150

    This change makes reversed bounded history listings keep the newest requested window and merely print that window oldest-first. It applies the probe-aware truncation consistently to log, author grouping, and revs, with regression coverage for skip, dates, JSON truncation, and OID order.

    Release grit log --reverse -n 100 median (100 runs; system Git 2.253 ms): 4.5 ms → 4.5 ms (0% worse)
  • ›GIT_INDEX_FILE ignored on write: grit commit/reset/rm --cached update the real .git/index instead of the temp index#2353f371cbd3

    Repository-level index writes now consistently follow the index selected by `GIT_INDEX_FILE`, including lock-based updates and commit-hook environments. This prevents commits, mixed resets, and cached removals from silently changing `.git/index` when an alternate index is active.

  • ›grit log --since combined with --author/--grep/pathspec walks all history: 1.3 s vs git 80-96 ms on git/git#2230f5729790

    Uses the existing date-bounded dense path walker when a commit filter is combined with a pathspec, then applies `--since`/`--until` eligibility before counting or emitting visible matches. The round-2 follow-up also keeps the required frontier parent available and deletes only an older grandparent in the missing-object regression.

    git/git --since + --author=Junio + Documentation median (7 runs): 1,085 ms → 58.3 ms (19× better)git/git --since + --grep=fix + Documentation median (7 runs): 1,005 ms → 49.9 ms (20× better)System Git author/path comparison median: 75.5 ms (system git vs review branch) → 58.3 ms (system git vs review branch) (30% better)System Git grep/path comparison median: 75.6 ms (system git vs review branch) → 49.9 ms (system git vs review branch) (52% better)
  • ›grit mv deletes the emptied source directory (git keeps it); running it from inside leaves the shell in a deleted cwd#235834f2b6fe

    Moves now rename tracked paths without deleting source directories that become empty. This also applies consistently to force-destination cleanup and rollback, preventing a move from invalidating a shell whose current directory was the emptied source.

  • ›Ref advertisement for 1M refs costs ~420-540 MB per request and 3x git's time (serve + ssh upload-pack), no streaming#2348e7f4d9c1

    Packed-refs iteration now streams sorted files while preserving the ref store’s sorted/unique callback contract. The round-3 fix first validates ordering without invoking the visitor, so a file that falsely claims `sorted` falls back to one sorted materialized emission instead of replaying records.

  • ›rebase --continue after a conflict in a multi-commit rebase prints 'Already up to date' and drops the rebase#2357f9ad748f

    This change makes multi-commit rebase resume consume the saved tail even when no explicit commit edits remain. Continue and skip now replay descendants from the seeded parent mapping, update the branch tip, reattach HEAD, and only then remove rebase state.

  • ›':(top)' pathspec magic fails from a subdirectory in diff/add/restore/rm/stash push/status (':/' form works)#2355247fe19f

    Long-form `:(top)` pathspecs now remain anchored at the repository root when grit is invoked from a subdirectory. The change covers worktree and revision-walk resolution and verifies diff, status, add, restore, rm, stash, range diff, glob, and exclude behavior against system Git.

  • ›merge dir/file conflict overwrites the directory; --abort leaves tree dirty; grit commit then drops the dir#2359101d0590

    This change relocates the file side of a directory/file conflict to a labeled sibling such as `f~side` instead of writing it over the directory. It also makes abort remove blocking files before restoring the original tree, while preserving correct unmerged index stages for staged and auto-stage commit behavior.

  • ›log/revs with ':(attr:...)' pathspec return nothing ('No commits yet.'); add ':(attr:...)' exits 0 and stages nothing#2356e583fb8e

    Attribute pathspecs now use the repository attribute stack in path-limited history and staging, including patterns ending in a directory slash. Empty filtered logs are silent instead of claiming the repository has no commits, while truly empty repositories retain that message.

    Debug attr-log median latency (100 runs): 6.9 ms → 8.2 ms (16% worse)
  • ›grit add/commit silently records a nested git repo as an empty gitlink: no warning, no hint, clones lose it#2360cb2cd546

    Nested repositories staged by add or commit are still recorded as Git-compatible mode-160000 gitlinks, but grit now exposes that fact through library outcomes and warns users that clones will not contain the nested files. Status identifies embedded repositories, clone distinguishes real submodules from bare gitlinks, and the round-two fix makes commit dry-run work for newly staged gitlinks instead of failing internally.

  • ›fetch killed by TERM/HUP leaves tmp_pack_* / .pack-*-install-* in objects/pack; grit gc never reclaims them#2362d6319487

    Fetch and pull now unwind through library cancellation on TERM/HUP, clean their in-progress pack artifacts, and return exit 130. Gc and prune reclaim expired temporary object artifacts while preserving fresh, disabled-by-policy, dry-run, unrelated dotted paths, and install-shaped regular files.

  • ›rebase --skip always undoes the whole rebase ('Already up to date'); 3 rebase_conflicts_cli tests fail on main#2363d6319487

    Resume planning now distinguishes a truly empty rewrite from a conflict session that still carries a seed or replay tail. Skip can drop the stopped commit and continue can preserve the user’s resolution while replaying descendants; human output reports branch-tip movement when no rewritten rows are displayed.

  • ›status/commit abort with pathless 'I/O error: Permission denied' when a directory is unreadable; git warns and goes on#2361b19e946f

    Unreadable worktree directories are now non-fatal during status and commit. Grit emits structured, pathful diagnostics, keeps inaccessible tracked entries unchanged, continues reporting accessible changes, and exposes warnings in status JSON.

  • ›upload-pack / branch listing on a files-backend repo with 1M packed refs: 2.0 s, 336 MB (git 0.16 s, 66 MB)#2351183c164c

    Streams mmap-backed packed refs into ref enumeration and merges them with loose refs, carrying packed peel information and respecting callback early termination. The round-two patch tightens line validation and delays emission by one record to address corrupt input and lying `sorted` headers, but delayed emission is insufficient for a later inversion.

    grit branch -n 1 median wall time on 1M packed refs: 1.4 seconds → 0.017 seconds (82× better)grit branch -n 1 peak RSS on 1M packed refs: 335,512 KB → 74,592 KB (4.5× better)upload-pack wall time: origin/main versus reviewed branch lower bound: 2.1 seconds (after is incomplete lower bound) → 732 seconds (after is incomplete lower bound) (100% worse)
  • ›grit commit/add of a nested repo with no commits (e.g. `cargo new`) records an empty 100755 file; status never clean#2364aed84ba9

    Embedded repositories whose HEAD does not resolve can no longer be turned into empty executable blobs or used to collapse unmerged gitlink stages. Staging and commit return a typed error without changing the index, while normal status output collapses nested untracked repositories consistently.

    Conflict-stage rows remaining after add: 1 rows → 3 rows (3.0× better)
  • ›Follow-up: Fix #2351: upload-pack / branch listing on a files-backend repo with 1M packed refs: 2.0 s, 336 MB (git 0.16 s, 66 MB)#23518ccac076

    Streams sorted packed refs without retaining a second copy of every ref name, while preserving sorted, unique loose-over-packed listing semantics. A validation pass detects lying sorted headers before callbacks, then either streams the mmap-backed records or materializes a sorted fallback.

    1M-ref advertisement median time: origin/main to reviewed branch: 0.493 seconds → 0.490 seconds (1% better)1M-ref advertisement median peak RSS: origin/main to reviewed branch: 67,596 KB → 67,684 KB (0% worse)1M-ref advertisement median time: system Git to reviewed branch: 0.162 seconds → 0.490 seconds (67% worse)1M-ref advertisement median peak RSS: system Git to reviewed branch: 63,220 KB → 67,684 KB (7% worse)
  • ›merge resolves adjacent edits cleanly (delete block vs insert right after it) where git reports a conflict#23697698f6d0

    This changes one-sided three-way merge span handling so an insertion touching a deleted or replaced ancestor span is grouped into the same conflict. It also fixes the retained-side marker body for replace-plus-trailing-insert conflicts and adds system-Git byte-for-byte regression coverage.

  • ›grit-lib docs: CommitRequest { .., ..Default::default() } snippet does not compile (private fields); other structs same#23746cc3f2f5

    Makes the repository identity guide compile-backed and usable by external grit-lib callers. The guide now demonstrates CommitRequest::new, assigns the resolved author through the public field, and explains that an unset committer is resolved from the repository.

  • ›switch/merge refuse 'uncommitted changes' on clean tree when HEAD tree has legacy 100664 modes (git/git history)#2372882b90b2

    This change makes tree transitions treat legacy regular-file modes as their canonical index equivalents. Clean repositories with historical 100664/100775 tree entries can now switch or fast-forward merge, while actual content and executable-bit changes still block as appropriate.

  • ›blame has no moved-line (-M) detection: -C levels don't imply it, moved blocks blamed to the moving commit#23707bd852a7

    Adds opt-in within-file moved-line attribution to the library and exposes it as `grit blame -M`; all copy-detection levels imply move detection to match Git. The round-two fix corrects the small permutation oracle so its expected commit ownership now matches both system Git and grit.

    30-line fixture lines attributed to original commit: 20 lines (system Git: 30) → 30 lines (system Git: 30) (50% better)
  • ›main regression: grit push --tags overwrites remote annotated tags with lightweight ones (peeled oid)#236631ac3c15

    This change prevents packed annotated tags from being treated as their peeled commit when refs are listed. Push planning now retains the tag-object OID, so `push --tags` reports an existing identical tag as up to date instead of silently replacing the remote annotation with a lightweight tag.

  • ›grit-lib: DiffOptions::renames ignored by build_file_diffs; documented diffstat recipe over-counts vs git diff#2371dc8def8d

    This makes the documented tree-diff-to-diffstat library path honor rename options, so renamed content is paired instead of counted as a deletion plus an addition. It also makes `DiffTarget::Trees` consistently accept commit or tree object IDs, including the empty-left-side/root-tree case fixed in review round 2.

  • ›pick --abort / --skip and reset --hard leave the conflicted file behind as untracked (modify/delete conflict)#2367d8a18b56

    Hard reset now removes worktree files represented only by unmerged index stages when those paths are absent from the target, fixing leaked conflict files after pick/revert abort, skip, and reset --hard. The round-2 safety fix prevents that cleanup from traversing a symlinked parent and deleting data outside the repository.

  • ›grit rebase of merge history duplicates commits: a topic forked from an earlier merge keeps its old ancestry#23650bdfd473

    Merge-preserving rebases now use a stable priority topological order, so every in-range parent is rewritten before its child while ready side branches still precede ready mainline commits. This prevents a topic forked after an earlier merge from retaining old ancestry and duplicating that earlier topic and merge in the rebased graph.

    Commits ahead after issue reproduction: 6 commits → 4 commits (50% better)t1 subject occurrences: 2 commits → 1 commits (2.0× better)merge t1 subject occurrences: 2 commits → 1 commits (2.0× better)Tiny-fixture grit rebase wall-time median: 90.8 ms over 20 measured runs after 3 warmups → 89.9 ms over 20 measured runs after 3 warmups (1% better)
  • ›kill -9 during grit commit leaves .git/next-index-<pid> (a full index copy) that no later command or gc ever removes#237779c4838e

    Interrupted commits can no longer leave full index copies behind indefinitely. Index-locking operations and gc now identify `next-index-<decimal pid>` files whose owner process is gone, while retaining live-process, current-process, and malformed-name files.

  • ›grit mv Dir dir (case-only directory rename) fails on case-insensitive filesystems: tries to move Dir into dir/Dir#2376ce4b321c

    This change makes a directory rename that only changes letter case resolve to the requested destination spelling instead of treating that spelling as an existing directory and appending the source basename. It also preserves unrelated destination collision protection and avoids treating symlink pairs as the same directory dentry.

  • ›grit merge/pick/rebase abort with stack overflow (infinite recursion) when conflicted dir `a/` has sibling dir `a0/`#2381c4b459e0

    Prevents merge-tree materialization from treating a sibling such as `a0/` as a descendant of `a/`, which previously caused unbounded recursion and SIGABRT. Invalid prefix relationships now return an error, and partial writes descend from an empty tree when a selected directory is absent from the base.

    five-file baseline grit merge latency before abort vs fixed conflict return: 0.042 seconds → 0.021 seconds (95% better)merge process exit status (SIGABRT shell code vs expected conflict code): 134 exit code → 1 exit code (134× better)system Git vs fixed grit five-file conflict latency: 0.009 seconds → 0.021 seconds (56% worse)
  • ›grit reset --hard over a 20k-50k file diff is 5x slower than grit switch and 1.7-2.5x slower than git, at 2.7x git's RSS#23757660278e

    Hard reset now plans rollback from the pre-alignment index, avoids per-path cache-tree invalidation on wide updates, attaches the target cache tree during checkout, and skips a redundant post-checkout stat pass when bulk checkout populated every dense entry. The round-3 follow-up confines spool instrumentation to unit tests and replaces the banned unit mutex with a bool-backed state lock.

    20k-file reset median elapsed (system git reference: 0.601s): 0.884 seconds → 0.848 seconds (4% better)20k-file reset median peak RSS (system git reference: 11,860 KiB): 37,640 KiB → 37,822 KiB (0% worse)5k-file reset median elapsed (system git reference: 0.101s): 0.221 seconds → 0.206 seconds (7% better)
  • ›grit reset --hard / merge --abort rewriting 30-60k files is 2.8-3.3x slower than git (and 5x slower than grit switch)#23110f338091

    Hard reset can now parallelize its wide worktree stat scan while keeping index preload disabled, preserving the intended memory bound. The branch also carries the previously reviewed merge-abort simplification and result-tree checkout wiring; all changes remain internal to grit-lib.

    100k-file delete/add hard reset median: origin/main to branch: 0.198 seconds → 0.108 seconds (84% better)100k-file delete/add hard reset: system git to branch: 0.218 seconds → 0.108 seconds (2.0× better)100k-file branch reset versus branch switch: 0.465 seconds → 0.108 seconds (4.3× better)
  • ›grit log A..B -n 5 -- <path> costs 1.6-8 s on linux regardless of N and range size (git: 0.09-0.14 s)#2383fffe90f9

    This replaces eager closure of all excluded history with a two-sided range paint and a compact hide predicate. Bounded range walks with path, author, parent, topo, stream, group-by, and dense-path filtering can now stop near the common frontier instead of parsing the entire hidden ancestry.

    50,021-commit synthetic range+path warm wall time (debug build; system Git reference 2 ms): 1,518 ms median of 5 → 9 ms median of 5 (169× better)50,021-commit synthetic range+path peak RSS: 85.5 MiB median of 5 → 47 MiB median of 5 (82% better)
  • ›Restoring a 400 MB incompressible blob takes 1.3-2.1 s vs git 0.5 s: streaming inflate is CPU-bound#2379f31fb840

    This change keeps large-object reads and writes bounded in memory while reducing zlib and pack-mapping overhead. It bounds inflate windows, streams CRC verification during packed reads, streams writes from 1 MiB, and chooses stored zlib for first chunks that do not compress; round two repairs truncation handling and restores geometric loose-buffer growth.

    64 MiB loose restore median (grit): 95.7 ms → 99.5 ms (4% worse)64 MiB loose grit/git ratio: 1.7 ratio → 1.8 ratio (7% worse)64 MiB packed restore median (grit): 103.9 ms → 101.5 ms (2% better)64 MiB packed grit/git ratio: 1.2 ratio → 1.1 ratio (4% better)
  • ›grit revs --since DATE walks the whole history: linux 31-70 s vs git rev-list 0.09 s (log --since is fine)#2382b2ba280f

    This change makes an unbounded default-order `--since` revision walk stop at the date frontier instead of traversing the entire reachable history. The review follow-up restores correctness for parent selectors and for `--count --skip --since`, with direct comparisons against system Git.

    100k-commit synthetic history, 100 selected: origin/main grit vs review branch median: 180.2 ms → 2.5 ms (72× better)100k-commit synthetic history, 100 selected: system Git vs review branch median: 1.6 ms → 2.5 ms (35% worse)
  • ›core.fsmonitor never speeds up grit status: entries never marked valid (38 ms vs git 21 ms on 20k files)#23780cdd8bcd

    This change teaches status and tracked staging to persist fsmonitor-valid bits for files a real worktree scan proved unchanged. The round-3 correction keeps chmod/type-only changes invalid and visible on subsequent status calls while retaining the large-repository speedup.

    20k-file grit status mean (latest origin/main → branch): 5.6 ms → 5.7 ms (2% worse)20k-file status mean (system Git → branch grit): 5.3 ms → 5.7 ms (8% worse)
  • ›grit clone with credentials in the URL fails to download the server's clone bundle (auth lost), falls back to full fetch#238479516944

    The HTTP client now turns complete percent-decoded credentials embedded in a clone URL into an in-memory, origin-scoped Basic authorization cache. Same-origin bundle downloads reuse that cache without putting secrets in the advertised URI or resumable state; rejected URL credentials fail immediately without invoking a helper, and resumable auth failures no longer consume five transport retries.

  • ›grit switch/reset --hard to a commit that deletes K and adds K files: 5-12x slower than git, 11x user CPU#2380358314ed

    This change replaces the mixed checkout's broad untracked walk with targeted collision checks and applies hard-reset index changes in bulk. Round 3 adds a restricted pre-reset worktree diff so unchanged paths avoid a second full stat scan while ordinary dirty paths are still restored.

    100k/K=10k switch wall time: system Git to reviewed grit: 0.588 seconds → 0.931 seconds (37% worse)100k/K=10k real reset wall time: system Git to reviewed grit: 0.692 seconds → 1.0 seconds (33% worse)
  • ›revs A..B / RevWalk::commits list commits reachable from A when dates are skewed (git/git v2.42.0..v2.42.1: 150 vs 145)#2373e4d868f4

    Repairs no-commit-graph range walks when committer-date skew lets an excluded-side path reach commits already popped as positive-only. The repair now traverses painted frontier state first, bounds unpainted bridge exploration, and keeps both list and count APIs aligned with system Git.

    v2.42.0..v2.42.1 optimized median: 0.066 seconds → 0.127 seconds (48% worse)v2.3.0..v2.3.1 optimized median: 0.116 seconds → 0.340 seconds (66% worse)v2.2.3..v2.3.9 optimized median: 0.117 seconds → 0.339 seconds (65% worse)v2.42.0..v2.42.1 count: 150 commits (Git: 145) → 145 commits (Git: 145) (3% better)
  • ›Follow-up: Fix #2380: grit switch/reset --hard to a commit that deletes K and adds K files: 5-12x slower than git, 11x user CPU#238001b4ce99

    Hard reset now overwrites staged content on paths whose committed tree entry is unchanged between the source and target commits. The performance test also measures a real reset on every Git sample instead of timing no-ops, and a one-line revwalk call-site repair keeps the rebased branch compiling.

    Mixed reset median: 1.2 seconds → 1.2 seconds (0% worse)Mixed reset ratio versus system Git: 1.5 ratio → 1.5 ratio (0% worse)
  • ›Follow-up: Fix #2373: revs A..B / RevWalk::commits list commits reachable from A when dates are skewed (git/git v2.42.0..v2.42.1: 150 vs 145)#237392dfecbf

    This repairs range-walk membership under severe committer-date skew while keeping the hide-side repair local. The follow-up makes the performance regression test discriminate the bounded implementation from the exact prior unrestricted traversal pattern that previously slipped through at the ceiling.

  • ›restore_workspace_snapshot does not undo a switch/commit: files from the undone state stay on disk as untracked#2385430cba12

    Restore now records a temporary tree representing files actually present on disk before refs and the index are rewound. Checkout compares that disk tree with the captured snapshot, so undo removes files introduced by a switch/commit and recreates tracked files deleted after capture while preserving the existing index and operation-state restoration sequence.

  • ›upload-pack stateless-rpc (v0/v1 smart HTTP) sends shallow list after ACKs: git fetch --deepen/--unshallow fails#23894ee672de

    Stateless protocol v0/v1 upload-pack now emits any deepen-driven `shallow`/`unshallow` records and their terminating flush before processing haves and writing ACKs. It retains existing stateful and protocol-v2 behavior and avoids inserting a leading flush for requests that contain only client `shallow` lines.

  • ›WorkspaceSnapshot cannot be persisted: no Serialize/encode API, and its worktree tree is unreferenced so gc prunes it#238712456b16

    Workspace snapshots can now be serialized and restored across processes, while explicit pin/unpin APIs keep their generated worktree trees and blobs alive through Git and grit garbage collection. The round-2 correction makes that persistence usable with non-self-describing binary serde formats by using an externally tagged RawRef enum and correctly encoding optional index mtimes.

  • ›merge: final-newline change on one side is lost when that side also adds/removes lines earlier (git keeps it)#2392e61d54bf

    Default three-way text merges now treat a missing final newline as a real line-record change. This preserves the newline state selected by the side that changed it, even when that side also inserted or deleted earlier lines, and removes an unsafe length-dependent workaround.

  • ›Custom RefStore with default rewrite_reflog loses reflog entries under reflog expire; refstore_conformance passes it#23953fb2620b

    Custom reference backends can no longer inherit a racy reflog rewrite that silently overwrites concurrent appends. Each backend must implement a locked read-transform-write, and the shared conformance suite now exercises serialization against a concurrent ref update.

  • ›Repository::fsck reachability roots ignore an injected RefStore's reflogs: reflog-only commits are reported dangling#239605a49e60

    Fsck now treats commits referenced only by an injected ref backend's reflogs as reachable, preventing embedders from misclassifying undo history as dangling. The same backend-aware reflog access is used for rev-list, while linked-worktree HEAD reflogs continue to be collected from their administrative directories.

  • ›clone_into with an injected RefStore writes HEAD and branch reflogs to <git_dir>/logs; HEAD@{0} then fails#2393a8fc2125

    Clone setup now writes every clone reflog entry through the repository's configured RefStore. Embedders using MemoryRefStore or another injected backend receive HEAD and branch history without stray `<git_dir>/logs` files, while files-backed repositories retain the existing Git-compatible on-disk path.

other changes34
  • ›Quality #67: Move logic into grit-lib49c17c71

    Moves uncommitted blame identity selection and hunk line-range ownership from grit-cli into grit-lib. Library users can now obtain hunk line slices and the same uncommitted author labels as the CLI without reimplementing cursor-based regrouping.

    blame-deep-2000 grit median: 22.5 ms → 23.0 ms (2% worse)blame-large-tail-full grit median: 30.6 ms → 30.9 ms (1% worse)blame-large-mid-churn-full grit median: 10.8 ms → 10.9 ms (0% worse)
  • ›Speed: switch/sparse@sparse-unique-120kfcd91f91

    Sparse switches now leave absent out-of-cone directories opaque when their index row still represents HEAD, retargeting the row directly to the destination subtree instead of expanding tens of thousands of descendants. The checkout also rebuilds invalid sparse cache-tree roots and avoids reloading the full index for collision reporting unless `core.ignorecase` is enabled.

    switch/sparse@sparse-unique-120k median wall: 352.1 ms → 22.0 ms (16× better)switch/sparse@sparse-unique-120k peak RSS: 52.1 MiB → 17.6 MiB (3.0× better)switch/sparse@sparse-120k median wall: 98.0 ms → 21.3 ms (4.6× better)switch/sparse@sparse-120k peak RSS: 35.1 MiB → 17.1 MiB (2.1× better)
  • ›Main keeper: make CI on main green (public-api)fcd91f91

    This branch contains no changes: it points to the same fcd91f911 commit as origin/main. The intended IndexHandle public-API snapshot ordering correction is already present on main from 49c17c714, so integration is a no-op.

  • ›Speed: ignore/check-1000@synth-10k59ff522a

    This change caches per-directory ignore state so bulk ignore checks no longer rebuild ancestor rule stacks or allocate matching groups for every path. It also normalizes CLI path batches once, adds coverage, and records the benchmark result; round two only rewrote the four commit messages, leaving every tree unchanged.

    ignore/check-1000@synth-10k median wall: 7.1 ms → 5.9 ms (19% better)ignore/check-1000@synth-10k grit/git time ratio: 3.1 x → 2.4 x (31% better)ignore/check-1000@synth-10k peak RSS: 18,141,184 bytes → 18,472,960 bytes (2% worse)
  • ›Quality #62: Deduplicate6ebf5fbe

    Rebase sessions now reuse the shared history commit readers instead of maintaining private copies. Corrupt object IDs in saved rebase plans return a specific public error, with a clear CLI message and regression coverage that verifies the operation state is retained.

    rebase/onto@synth-10k median wall time: 12.6 ms → 12.6 ms (1% worse)rebase/onto@synth-10k peak RSS: 16,936,960 bytes → 16,756,736 bytes (1% better)
  • ›Quality #69: Simplify4466dfc0

    Deletes the unused rev-list pretty/format renderer and replaces its remaining parent-walk signal with the existing explicit `parent_rewrite` option. Supported `grit log` and `grit revs` rendering remains separate and unchanged; tests now validate the retained parent-rewrite behavior directly against system Git.

    log/oneline-100@git-git grit median wall: 3.9 ms → 3.5 ms (14% better)log/oneline-100@git-git system Git control: 3.1 ms → 2.8 ms (10% better)revs/count-path@git-git grit median wall: 387.8 ms → 386.1 ms (0% better)revs/count-path@git-git system Git control: 399.1 ms → 397.3 ms (0% better)
  • ›Quality #66: Dead code and dependencies4639b7ec

    Removes dead and test-only grit-lib APIs, an unused reflog formatter and history counters, and stale benchmark fixture code. It updates the public API snapshot and library layout documentation, while repairing benchmark target compilation and warnings. Round 2 only rewrites the two commit messages to remove forbidden co-author trailers; the source trees are unchanged.

  • ›Quality #63: Simplify012c2b31

    Removes one unused private reflog formatter and deletes an unused 10k-file benchmark fixture. It also resolves benchmark-only unused-result, unused-field, and unused-import warnings without changing benchmark bodies or production behavior.

    Lines across the six changed files: 3,677 lines → 3,627 lines (1% better)Net changed-area source size: 60 changed lines retained versus replaced/deleted → 10 changed lines retained versus replaced/deleted (6.0× better)
  • ›Speed: branch/merged@many-refs7ce32b6d

    The ref-query hot path now records each resolved peel once, filters rows with a keep mask, and avoids unnecessary commit-cache/ranking work when all tips are merge targets. It adds Git-compatibility coverage and committed benchmark evidence; round two only rewrites the two commit messages, leaving both code trees byte-for-byte unchanged.

    branch/merged@many-refs median, system Git 1.372 ms (20 runs): 2.7 ms → 2.4 ms (11% better)branch/merged@many-refs peak RSS (20 runs): 17,520 KiB → 16,320 KiB (7% better)branch -a --merged main median, system Git 25.155 ms (100 interleaved runs): 14.1 ms → 13.2 ms (6% better)branch -a --no-merged main median, system Git 9.845 ms (100 interleaved runs): 11.0 ms → 10.8 ms (2% better)
  • ›Quality #68: Deduplicatec2a0db38

    Adds one canonical way to obtain the tree named by HEAD and replaces repeated resolve/read/parse sequences across diff and porcelain paths. Strict callers use `Repository::head_tree`; callers that historically treated invalid HEAD like an empty repository use `HeadState::tree`.

    add/modified@synth-10k (system git 17.21 ms): 12.3 ms median → 11.4 ms median (8% better)commit/paths@synth-10k reverse-order 30-run repeat (system git 21.83 ms): 23.0 ms median → 23.1 ms median (1% worse)diff/worktree@synth-10k paired run (system git 7.42 ms): 7.6 ms median → 7.4 ms median (3% better)mv/file@synth-10k reverse-order 30-run repeat (system git 5.65 ms): 9.4 ms median → 9.4 ms median (1% better)
  • ›Speed: config/set@small42997e6e

    Adds a lightweight repository locator for config-only operations so `grit config` can avoid opening refs and the object database. The locator reuses one merged config cascade, preserves full discovery semantics for layout-changing configuration, and config writes retain lockfile-plus-rename atomicity while dropping an unnecessary fsync.

    config/set@small grit wall: 2.9 ms median (20 runs) → 1.8 ms median (20 runs) (58% better)config/get@small grit wall: 2.1 ms median (20 runs) → 1.8 ms median (20 runs) (13% better)config/get-origin@small grit wall: 1.9 ms median (20 runs) → 1.8 ms median (20 runs) (9% better)config/set@small grit peak RSS: 11.5 MiB median peak RSS → 10.9 MiB median peak RSS (6% better)
  • ›Speed: pick/one@synth-10k0331bcdc

    Index serialization now streams through a bounded 128 KiB hashing buffer directly into the lock file instead of retaining a complete serialized index body in memory. The review-fix round only rewrote the three commit messages; its tip is tree-identical to the already reviewed implementation.

    pick/one@synth-100k interleaved median wall: 152.4 ms → 149.0 ms (2% better)pick/one@synth-100k interleaved peak RSS: 46.0 MiB → 38.3 MiB (20% better)pick/one@synth-10k median wall: 56.3 ms → 55.9 ms (1% better)pick/one@synth-10k peak RSS: 21.7 MiB → 21.6 MiB (1% better)
  • ›Quality #64: Library API qualityab803673

    This change replaces a string-based corrupt rebase-plan OID failure with a documented `RebaseSessionError::InvalidStateOid`, including the offending value. It also deduplicates rebase-session commit reads onto shared history helpers while preserving each caller's established typed outer error.

    rebase/onto@synth-10k median wall time: 13.2 ms (origin/main vs branch, 20 runs each) → 13.2 ms (origin/main vs branch, 20 runs each) (0% worse)rebase/onto@synth-10k peak RSS: 16,920,576 bytes (origin/main vs branch) → 16,998,400 bytes (origin/main vs branch) (0% worse)
  • ›Main keeper: make CI on main green (scripts, test)359e971b

    This change repairs the files-ref backend’s concurrent file/directory create handling and restores the scripts CI mapping check. The round-2 correction preserves Git-compatible behavior by replacing a stale empty directory at the exact ref path, while directories containing refs or in-flight locks remain conflicts.

  • ›Speed: diff/worktree@synth-10k539f7d04

    This change lowers worktree-diff memory by shrinking stale-stat hash jobs to index positions plus exceptional attributes, then constructing diff entries only for changed files. It also replaces unbounded per-directory merged attribute caching with a shared stack for the current directory chain, with regression tests and committed before/after benchmark reports.

    Target peak RSS, exact branch base to tip: 29.8 MB → 22.2 MB (34% better)Target median time, exact branch base to tip: 7.0 ms → 6.8 ms (3% better)Target time ratio versus system Git: 1.1 x Git → 1.1 x Git (3% better)
  • ›Quality #65: Move logic into grit-libc0a7d261

    Blame results now carry complete placeholder metadata for uncommitted worktree or caller-buffer lines, and each library hunk records both its uncommitted state and the range of attributed lines it covers. The CLI no longer invents that identity or walks hunks with its own cursor; it only renders the library result.

    Plain blame median, 300 commits / 3,000 lines / worktree edit: 67.8 ms → 69.2 ms (2% worse)Grouped JSON blame median, same fixture: 67.7 ms → 69.1 ms (2% worse)
  • ›Main keeper: make CI on main green (scripts, test)ad3fc729

    Repairs the upload-pack advertisement compile break introduced by the streaming request-body merge and synchronizes the generated public-API snapshot. The Rust change supplies an empty mutable reader without changing advertised bytes or protocol behavior; the snapshot change only reorders existing API entries.

  • ›Main keeper: make CI on main green (clippy, test, rustdoc, public-api)ea496e8c

    Fixes the docs site’s strict validation by removing a rustdoc link to a doc-hidden FetchOptions method and correcting the switch page’s relative link to the clone command. The delta is two one-line documentation corrections and does not alter Rust behavior or public APIs.

  • ›Perf regression 2026-10-27: commit/staged@synth-100k, diff/staged@deep-tree, diff/worktree@deep-tree +445a1c367

    Hard reset now stats tracked files relative to open directory handles instead of repeatedly walking absolute paths, and it uses a sorted stat vector rather than a hash map. Checkout-created regular files are marked content-verified so index writing does not immediately reread/re-hash them; size-changed regular files are rewritten directly.

    reset/hard@synth-10k grit wall (current main → branch): 20.4 ms → 18.3 ms (12% better)reset/hard@synth-10k system Git control: 15.1 ms → 16.0 ms (6% worse)reset/hard@synth-10k grit/Git ratio: 1.4 ratio → 1.1 ratio (18% better)reset/hard@synth-10k peak RSS: 17,976 KiB → 18,708 KiB (4% worse)
  • ›Speed: status/dirty@synth-10ke41aa10f

    Adds an opt-in grit-lib hot-text layout that groups frequently co-executed library/runtime/dependency functions into a contiguous ELF section on x86_64 GNU/Linux. The grit CLI enables it by default, cutting short-command code-page RSS while leaving repository behavior and formats unchanged.

    status/dirty@synth-10k wall ms (origin/main → branch; git 11.248 ms): 7.0 ms → 7.2 ms (2% worse)status/dirty@synth-10k peak RSS (origin/main → branch; git 9.83 MB): 17.0 MB → 14.2 MB (19% better)status/clean@synth-10k wall ms (same-code layout off → on): 7.0 ms → 7.0 ms (0% worse)status/clean@synth-10k peak RSS (same-code layout off → on): 16.7 MB → 17.4 MB (4% worse)
  • ›Quality #71: Move logic into grit-lib2188ced3

    Branch creation now returns the branch's effective upstream as part of the library outcome. The CLI simply renders that field, eliminating its separate dry-run tracking resolver and post-create config lookup; forced-reset dry runs now match live output when existing tracking is retained.

    branch/create@many-refs median wall time (30 runs): 2.6 ms → 2.5 ms (3% better)branch/create@many-refs peak RSS: 13,783,040 bytes → 14,008,320 bytes (2% worse)system git control median wall time: 1.6 ms → 1.6 ms (1% better)
  • ›Quality #70: Library API quality9f24c3ed

    Remote-supplied Git protocol failures now remain a typed `Error::RemoteFatal` instead of being flattened into generic message errors. Round 2 closes the smart-HTTP discovery gap so v0 and v2 `ERR` responses fail consistently through both public body-discovery APIs.

    ls-remote/all@many-refs median wall: 14.1 ms → 13.9 ms (2% better)ls-remote/all@many-refs peak RSS: 19,374,080 bytes → 19,673,088 bytes (2% worse)system git control median wall: 15.4 ms → 14.7 ms (5% better)
  • ›Main keeper: make CI on main green (test)a477324a

    This change detects an explicit non-LLD linker in Cargo's encoded or plain Rust flags and turns off grit-lib's LLD-only hot-text linker-script injection for that build. It also gates the grit-cli ELF-layout assertion under the same decision and adds regression coverage for linker detection.

  • ›Speed: reset/undo-commit@git-gitf30cdeb5

    Reset now avoids reading packed ref tips when their file timestamp proves they cannot reach newly orphaned commits, while retaining a full-walk fallback. It also moves the ref with compare-and-swap and appends reflogs directly, avoiding transactions whose cost grows with reflog length.

    reset/undo-commit fresh median wall: 7.8 ms → 7.2 ms (8% better)reset/undo-commit fresh peak RSS: 23.7 MiB → 24.3 MiB (3% worse)reset/undo-commit aged packed-refs median wall: 7.5 ms → 3.8 ms (2.0× better)reset/undo-commit aged packed-refs median peak RSS: 23.6 MiB → 14.8 MiB (59% better)
  • ›Quality #72: Dead code and dependencies2bb30403

    Removes test-only rev-list APIs left behind by the retired compatibility CLI and deletes the now-unreferenced duplicate ref-exclusion implementation. It also drops an unused HTTP-server dependency and an unused reftable test accessor, for a net reduction of 832 lines while preserving production walk and ref-advertisement behavior.

    revs --all --count median (3k commits, 20,001 refs): 17.3 ms → 17.1 ms (1% better)log --all -n 100 median (3k commits, 20,001 refs): 7.9 ms → 7.9 ms (1% worse)
  • ›Main keeper: make CI on main green (public-api, test)8fd76f3d

    Refreshes the checked-in grit-lib public API snapshot so it matches the API already emitted by the library. This is metadata-only: no Rust implementation, test behavior, on-disk format, wire protocol, or CLI behavior changes.

  • ›Main keeper: make CI on main green (test, public-api)0c4fac54

    This corrects date-cutoff traversal so an older commit does not enqueue parents once every remaining queued commit is also older than `--since`. The same frontier ordering is applied to reachable, streaming, and path-limited walks, while bounded `--since` lists avoid a linear fast path that cannot preserve Git's frontier behavior.

    Date-limited log median (30 release runs, 12,300 commits): 4.0 ms → 4.1 ms (3% worse)
  • ›Main keeper: make CI on main green (test)208ff6ec

    This change stops reachable-history traversal at Git's `--since` frontier instead of following parents through an already-too-old commit. It also routes since-bounded linear histories through the general reachable-walk implementation so skewed dates receive the same frontier semantics.

  • ›Speed: branch/list@many-refsb5b09b22

    This change teaches the existing profile-driven linker-layout generator to emit a second hot-code fragment for CLI startup, clap, rendering, and CLI-instantiated generics. The shipped x86_64 GNU/LLD binary places that section beside grit-lib's hot section before ordinary text, reducing startup code faults and resident memory without changing command behavior or repository formats.

    branch/list time (system git 1.353 ms): 2.0 ms → 1.8 ms (12% better)branch/list peak RSS (system git 6.57 MiB): 10.1 MiB → 8.5 MiB (19% better)branch/list-json time: 1.9 ms → 1.9 ms (4% better)branch/list loose time: 1.9 ms → 1.7 ms (9% better)
  • ›Speed: branch/list-json@many-refs0bdfd473

    This review contains no code change. The investigation establishes that branch/list-json on many refs remains dominated by shared CLI startup costs, so the author correctly avoided a benchmark-specific grit-lib or CLI patch that would not materially improve the row.

    branch/list-json@many-refs grit median: 2.0 ms → 1.9 ms (2% better)branch/list@many-refs grit median: 1.8 ms → 1.8 ms (1% worse)branch/list@many-refs-loose grit median: 1.9 ms → 1.9 ms (3% better)branch/merged@many-refs grit median: 1.9 ms → 2.0 ms (3% worse)
  • ›Main keeper: make CI on main green (rustdoc, public-api)6def0af9

    This removes one redundant rustdoc link destination and updates one generated public-API snapshot field position. Together those two one-line corrections restore the rustdoc and public-api checks without changing executable behavior.

  • ›Main keeper: make CI on main green (public-api, rustdoc)0f338091

    This review branch contains no changes relative to origin/main. The two-line CI repair it references—one rustdoc link simplification and one public-API snapshot ordering update—was already merged into main as 6def0af921e6ba9d73044b1358165e8ae060c736.

  • ›Main keeper: make CI on main green (check-windows-msvc, rustdoc, test, clippy)bfa0a953

    This change restores compilation after the range-walk hide type refactor by supplying the expected empty `RangeWalkHide`. It also makes the public streaming-write documentation acceptable to rustdoc by referring to the private threshold constant as code rather than an intra-doc link.

  • ›Main keeper: make CI on main green (test, clippy, rustdoc, check-windows-msvc)01b4ce99

    Updates the since-cutoff streaming revision walk to pass the empty `RangeWalkHide` value required by `streaming_date_walk`. This is a one-line compilation repair with no behavior or compatibility surface change.

$ git log --oneline f91a862b..v0.5.27compare on GitHub ↗
a8fc2125Merge branch 'factory/fix-2393-clone-reflog-injected-store'
05a49e60Merge branch 'factory/fsck-injected-reflog-roots-6147'
3fb2620bMerge branch 'factory/rewrite-reflog-required'
e61d54bfMerge branch 'factory/merge-eof-newline-2392'
12456b16Merge branch 'factory/workspace-snapshot-persist-6123'
3a14cea1Add default-CI test for clone reflogs on injected RefStore
258c2c75Fix clone reflogs for injected RefStore
fc37f8eaFix fsck reachability roots for injected ref store reflogs
716afca7Fixes #2395: require RefStore::rewrite_reflog for custom backends
4ee672deFix stateless v0 upload-pack shallow section before ACKs for deepen
dc8de49aFix merge EOF newline when side also changes line count
413dc70fUse externally tagged RawRef serde for bincode persistence
d7aadd04Fixes #2387: persist workspace snapshots and pin worktree trees for gc
430cba12Fix workspace snapshot restore worktree diff base
92dfecbfStrengthen skew repair perf guard against cd3f334b regression
6ba5c0c9Make skew range paint perf guard exercise capped repair
01b4ce99Fix git reset baseline setup in mixed swap perf test
0cea1407Fix hard reset Keep on tree-stable paths with staged index content
bfa0a953Merge branch 'factory/fix-streaming-since-range-walk-hide-6104'
209887fdfix(odb): avoid private intra-doc link on write_stream
e185beeefix(rev-list): pass RangeWalkHide to since-cutoff streaming walk
e4d868f4Bound skew hide-paint repair to BOTH frontier and per-start cap
42f8f67aExtend hide-paint repair through unpainted skew paths
bdf8a057Fix range revwalk hide paint under committer-date skew
358314edchore: sync integration manifest after rebase onto main
29b1bdecperf(checkout): fast hard reset for tree-stable paths
033db787Address review: restore hard-reset stat probes and ignore rules
de66452cFix switch/reset --hard perf on mixed wide tree delete/add
79516944Merge branch 'factory/fix-2384-clone-url-bundle-auth'
0cdd8bcdMerge branch 'factory/fix-2378-fsmonitor-valid-bits'
9b22c2beFix clone URL credentials for same-origin bundle downloads
b2ba280fMerge branch 'factory/revs-since-frontier-6080'
2c4e2c08Mark fsmonitor-valid on parallel content-verified clean path
e239317cDo not mark fsmonitor-valid on content-only refresh
4fd2a6f4Fix fsmonitor valid bits after clean status scan
fffe90f9Merge branch 'factory/log-range-pathspec-perf-6081'
f31fb840Merge branch 'factory/fix-2379-streaming-inflate'
c6f0a2a0Fix --since regressions for parent filters and count skip
393956d7Fix rev-list --since to stop at date frontier without -n
0070689cFixes #2383: paint range hide for bounded rev-list walks
0f338091Merge branch 'factory/fix-reset-hard-perf-2311'
a1141945chore: rustfmt checkout_incompressible_blob_restore
97f29c5cFix pack truncate hang and restore loose buffer growth
0d0a8e0cFixes #2379: speed streaming inflate and large blob writes
c4b459e0Merge branch 'factory/fix-2381-merge-sibling-dirs'
7660278eScope checkout spool probe to unit tests only
58910e62Fix racy wide tree-diff rollback on cancelled hard reset
5b99c562Fixes #2375: speed hard reset tree-diff checkout like switch
95f6df97Fix hard-reset stat scan parallelism without index preload
fb9b71beFixes #2311: parallel hard-reset stat scan and faster merge abort
3cc64d96Fixes #2381: stop merge stack overflow for sibling dir prefixes
6def0af9Merge branch 'factory/ci-green-rustdoc-public-api'
ce4b321cMerge branch 'factory/mv-case-dir-2376'
7ee4c2dfKeep mv destination-directory detection case-sensitive
34f3aa28Do not treat symlink pairs as case-only directory mv
6cf8ebccFix case-only directory mv on ignorecase volumes
79c4838eReap orphaned next-index hook copies after killed commits
fd8575c8Fix CI rustdoc and public-api snapshot on main
0bdfd473Merge branch 'factory/rebase-merge-duplicate-2365'
d8a18b56Merge branch 'factory/fix-pick-abort-unmerged-cleanup-2367'
dc8def8dFix rustdoc intra-doc links for rename detection docs
07cba95eFix DiffTarget::Trees with old=None and tree OID on new side
45a4def5Fixes #2371: honor DiffOptions::renames in build_file_diffs
31ac3c15Merge branch 'factory/fix-2366-packed-tag-push'
6fbae6e7Fix #2365 regression test parent inspection and fmt
c619c920Fixes #2365: priority topo sort for merge-preserving rebase replay
4bdcd9cdSkip conflict-only hard-reset removal through symlink parents
176f708dFix hard reset leaving modify/delete conflict files in worktree
7bd852a7Merge branch 'factory/blame-move-detection'
882b90b2Merge branch 'factory/legacy-tree-mode-switch-2372'
6cc3f2f5Merge branch 'factory/fix-2374-commit-request-docs'
43226acdAddress review: drop in-module packed-refs fixture, sync override
147825baFix #2366: list_refs uses tag object for packed annotated tags
fd3d617fFixes #2370: blame within-file move detection (-M)
82ef4f52Fixes #2372: canonicalize legacy tree modes in switch/merge transitions
7698f6d0Fixes #2369: conflict when delete touches adjacent insert
0fca6585Fixes #2374: compile Identity guide via CommitRequest::new example
b5b09b22Document the CLI hot-text layout and record its daily rows
09d390d0Lay out the grit binary's own hot code next to grit-lib's
208ff6ecSkip linear-chain shortcut for --since bounded rev-list walks
0c4fac54Fix rev-list --since frontier to match Git on skewed dates
8ccac076Merge branch 'factory/packed-refs-stream-5994'
8fd76f3dMerge branch 'factory/ci-green-6000'
11206344Stream packed-refs for ref listing without retaining every name
4bf9e3a8Refresh grit-lib public-api snapshot for pack_receive and pathspec API
2bb30403Remove test-only rev_list API and the dead ref_exclusions module
c30c4f1aRemove unused mapped_arc test helper from reftable
86cbb716Drop unused http-body-util dependency from grit-http-server
f30cdeb5Document the undo-commit numbers with tags packed before the new work
1d50cddaAppend reset reflog lines instead of logging through transactions
f78d148aSkip packed tags the orphaned-commit report cannot need
aed84ba9Merge branch 'factory/fix-embedded-repo-no-commit'
a477324aSkip hot-text-layout when CI forces GNU bfd linker
013aeb42Fix tracked-scan path for unborn embedded gitlinks
e9378c23Refuse unmerged gitlink staging when nested HEAD is unborn
4a3ce997Fixes #2364: refuse embedded repos without commits
9f24c3edReport a server's ERR packet and side-band fatal as Error::RemoteFatal
183c164cMerge branch 'factory/packed-refs-stream-f0c2'
2188ced3Merge branch 'factory/branch-create-upstream'
30671854Fix sorted fallback duplicates and streaming scan cost
6c8a52c6Fix packed-refs streaming strictness and sorted fallback
93f04f92Stream packed-refs listing without materializing full maps
b19e946fFix remaining status/commit aborts on unreadable directories
1442fac1Fix unreadable directories aborting status and commit
b8818248Report the configured upstream from create_branch
d6319487Merge branch 'factory/fetch-interrupt-gc-2362'
838bd128Merge branch 'factory/rebase-skip-2363'
cb2cd546Merge branch 'factory/embedded-gitlink-warnings-5962'
eb995b07Apply rustfmt after rebase
e583fb8eMerge branch 'factory/attr-pathspec-log-add'
a13b6504Restrict gc install-stage pruning to .pack-*-install-* dirs
ca057fbaFix fetch interrupt temp pack cleanup and gc stale tmp pruning
e9ca77f9Fix rebase --skip and --continue after conflict when edits are empty
00b3989aFix commit --dry-run for newly staged gitlinks
8709359cWarn when nested repositories are staged as gitlinks
f94d0db0Fix attr pathspec in rev-list, log, and add
101d0590Fix file/directory merge conflicts (#2359)
e41aa10fMerge branch 'factory/status-hot-text-layout-5960'
247fe19fMerge branch 'factory/pathspec-top-subdir'
68106588Document the hot-text layout and record its census baselines
f9ad748fMerge branch 'factory/rebase-continue-multi-c444'
0080094dAdd grit-lib hot-text-layout feature to cut code-page RSS
1f65fbd7Require non-empty status output in :(top) CLI test
26891b77Fix CLI regression test for :(top) pathspec from subdirs
31800974Fixes #2355: honor :(top) pathspec magic from subdirectories
e7f4d9c1Merge branch 'factory/ref-ad-stream-5915'
34f2b6feMerge branch 'factory/mv-keep-empty-dirs'
36799d03Fix #2357: add CLI regression tests for multi-commit rebase continue
b093b233Probe packed-refs order before streaming callbacks
21aa09d5Fix packed-refs for_each_ref merge and cancellation
1f80bd32Stream packed-refs listing for ref advertisements
d3d04516Fixes #2358: grit mv leaves emptied source directories on disk
f5729790Fix #2230: skip pre-since commits on bounded path walk
3b22dad5Fix #2230: use bounded path walk for since plus filter
f371cbd3Merge branch 'factory/git-index-file-write'
375f5150Fix log and revs --reverse -n window truncation
8892acb2Merge branch 'factory/switch-typed-branch-errors-a20c'
67732b6cFix rebase --continue misleading "Already up to date" output
98cb3a9dFix rebase continue after conflict resolution with empty edits
9ac871ebFix GIT_INDEX_FILE ignored on index writes
c83221ffFix switch branch resolution to return typed BranchError
4a45220bFix rebase continue/skip when first replay commit has no edits
45a1c367Stat hard-reset paths relative to their directory
f81e19f4Mark files a checkout just wrote as content-verified
ea496e8cdocs: fix Site check after fetch-stream merge
ad3fc729Fix CI after fetch-stream-cap merge
c0a7d261Fix blame import after hunk API conflict resolution
f08d5dd2blame: complete uncommitted metadata in grit-lib
539f7d04bench: record diff/* daily rows before and after the stale-index memory fixes
90d85445worktree_rules: keep attribute stacks for the current directory chain only
fd97f254diff: queue stale-stat hash jobs by index position
78b24ef3Merge branch 'factory/fetch-stream-cap-5886'
359e971bFix review: allow Git-empty ref dirs, tighten D/F race errors
28fa1f33Fix CI: refstore D/F create race and upstream mapping row
c29081c3chore: extend public-api snapshot for RequestBodyLimitExceeded
8be4fafbfix: map protocol v2 fetch body limit to HTTP 413
d332ee07fix: stream large fetch bodies over smart HTTP (Fixes #2347)
fced6cf3Update grit-lib public-api snapshot for ReftableReader::from_mapped
c7a307f2Fix review blockers: mapped reftable readers, streaming info/refs
ae260549Fix #2348: Stream ref ads and mmap reftable stack cache
ab803673Report invalid object ids in a saved rebase plan as a typed error
bf98ffa2Reuse shared history helpers in rebase sessions
42997e6eMerge branch 'cursor/config-set-fast-path-e424'
c2a0db38Merge branch 'factory/head-tree'
12b037d7Speed up grit config set/get: locate the repo without opening it, drop config fsync
2365fa54Fix diff_patch_text test build: borrow index line through a fn
0331bcdcMerge branch 'cursor/pick-index-write-memory-422b'
aba07d06Refresh grit-lib public-api snapshot for Repository::head_tree
8b218670Use Repository::head_tree instead of private HEAD-tree helpers
db6c7f0aAdd Repository::head_tree and HeadState::tree
89aa4f76Record pick/one before/after baselines and the peak-memory ledger
55f64fcbName the index chunk sink type for clippy::type_complexity
28216e19Stream index writes through a bounded buffer instead of a full in-memory body
012c2b31Remove dead reflog formatter and fix library bench warnings
7ce32b6dMerge branch 'cursor/branch-merged-peel-once-0837'
4639b7ecMerge branch 'cursor/dead-code-round-b0c4'
94ef3815Record branch --merged many-refs profile and baselines
183e5ea4Peel each ref tip once in reachability-filtered listings
4466dfc0Remove the unused rev-list pretty-format renderer
6ebf5fbeMerge branch 'cursor/rebase-session-shared-helpers-d5b1'
ea1f52b8Remove test-only grit-lib functions found by the privatizing dead-code scan
0dbe3ca7Remove dead reflog formatter, unread fixup counter and stale bench code
8a7acaabUse shared history helpers in rebase_session and type invalid plan oids
4ca11f3fBuild CommitRequest via new() in rebase session tests
59ff522aDocument and baseline the per-directory ignore stack change
6fa237d0Use RepoPathNormalizer and move results into output in grit ignore check
b08ce7aaReuse the query path stat to skip per-directory file checks in ignore_check
e5cf8831Build ignore exclude stacks once per directory and match without per-path allocation
e46dc720Honor protocol.version on SSH fetch, list_refs, and remote refs
fcd91f91Merge branch 'factory/sparse-switch-unique'
3829818cDocument the sparse switch numbers for changed excluded trees
d1ce8254Keep changed excluded sparse directories opaque on switch
49c17c71Refresh grit-lib public-api snapshot for blame hunk lines
19cf6133Move blame hunk lines and uncommitted author naming into grit-lib
3c14b849Merge branch 'factory/fix-2345-chunk-bounds'
34a2249eFixes #2343: write remote-tracking reflog on fetch
e5c71e7aFix repack chunking so same-path large blobs share one delta base
b704da15Fixes #2341: hash worktree content for patch-text index lines
a6ee9dd7Merge branch 'factory/status-untracked-cwd-paths'
7c3ec0caFix cwd-relative status paths for untracked directories
f8543ad2Fixes #2342: stop pairing untracked paths in unstaged rename detection
22dd4f2dFix stash list --json to return a list object