Timeline

LogStore.open now takes the exclusive lock before it reads the file's length and scans for the end, and a read-only handle refuses to append. purge undo opens the store through the shared writer path.

A new DST sweep runs two to four writers interleaved against one log through a file that models the lock. It flags unlocked or misplaced appends as they happen, checks each writer sees everything committed before it locked and reads back its own writes, including deltas. A source check fails on any write-mode open of the store outside openStoreAppend. An e2e test races commits, tags and user-table writes on one store and counts what landed. check-in: 5c445bb3f3 user: dmestas tags: trunk, version-0.3.32

It appended without the lock every other writer takes, so a concurrent commit, pull or push could overwrite its frames or have its own overwritten with valid frames nothing could detect. It now opens through openStoreAppend, which also resolves a detached worktree's store. Release 0.3.32. check-in: 2249f19056 user: dmestas tags: trunk

The artifact quota is shared with every repository on the account; a full one failed the log upload on two nights the soak passed. The log is now printed into the job output, which is kept 90 days outside the quota, and the upload no longer decides the run. check-in: 96b430bb17 user: dmestas tags: trunk

accept read the branch tip up to three times and merged it by name, so a push mid-accept could be merged unprobed. It now reads the tip once and merges that hash on both paths. --expect-tip HASH refuses with exit 6 and nothing written when the branch is no longer at HASH. Release 0.3.31. check-in: 60890d6661 user: dmestas tags: trunk, version-0.3.31

Every artifact it uploads is transit, published to the uv store and a GitHub release the day it is built. Thirty days kept ~40 MB per release against the 500 MB storage quota, and a full quota failed the soak's log upload. check-in: 29f07577ae user: dmestas tags: trunk

A Basic challenge was answered whatever the scheme, so over http the password crossed readable and the saved remote repeated it. It is now refused to non-loopback hosts, naming the https URL; a redirect to another host drops the credential; plain http to another host notes it on stderr. Release 0.3.30. check-in: b637cbd37c user: dmestas tags: trunk, version-0.3.30

A failing fuzz target is abandoned and the sweep moves on, so one recurring crash no longer hides the targets after it; the exit lists every finding. Each workflow clears its zig-local cache whole at job start once it passes 8 GB, since Zig never prunes it. check-in: 8ebc3e4dd8 user: dmestas tags: trunk

tossBitsShort added the consumed bits where it must subtract them, so a header ending one byte short tossed past the end of the input and the next peek read a wrapped length off the stack. Found by the nightly soak four nights running. Release 0.3.29. check-in: 362f18d357 user: dmestas tags: trunk, version-0.3.29

config export AREA [FILE] writes fossil's configuration export format, so an area can be backed up before import replaces it; import and merge now apply /shun cards, add-only in both modes. merge --probe with several NAMEs answers each against TARGET with the ones before it landed, holding merged files in memory and moving the merge base with the landing. Release 0.3.28. check-in: c5243e961f user: dmestas tags: trunk, version-0.3.28

The pusher's recovery from a refused push, amend --branch, was refused too. A cancel on an unheld target removes it from nothing held, so the check-in is judged where it joins, with a same-round branch move counted. A held check-in re-offered in a push is no longer judged as joining. The e2e raw-push fixture compared raw digests to hex and excluded nothing. Release 0.3.27. check-in: d4571a4b1a user: dmestas tags: trunk, version-0.3.27

The marker branch fell through to the operand slots, so it took VERSION's place and OUTPUTFILE was refused. A value option's argument is now also taken before the marker is considered. Release 0.3.26. check-in: 28cd7f506a user: dmestas tags: trunk, version-0.3.26

A bare -- ends the options in every verb; zip names are flagged UTF-8. check-in: 726efdca6b user: dmestas tags: trunk, version-0.3.25

A file whose name begins with '-' could not be named at all: passed bare it parses as a flag, and './-x.txt' parses and then matches nothing, because a path is compared literally against the check-in's file list. A forge serving such a repository answered 500 for every content route on that file. Twenty parsers already honoured the marker, including the shared flag scanner behind add, rm, commit and mv; the twenty-five with hand-written loops did not, cat among them. Each now gates its flag branches behind the marker, so operands fall through to the positional handling untouched and '-n' after '--' is a name rather than a limit. Documented as its own help topic rather than an options row in thirty-nine verbs: '--' is not a verb's option, it is the rule every verb parses by. check-in: 85a1d4bf3f user: dmestas tags: trunk

The name bytes were always correct UTF-8 and general-purpose bit 11 was never set, so a conforming reader must decode them as CP437 and a committed 'café.txt' extracted as 'caf├⌐.txt'. Both headers now take the flag from one constant: a reader trusts the central directory for metadata, so setting it on the local header alone would still ship the bug. Set unconditionally, since ASCII is a subset of UTF-8 and one value cannot disagree with itself. Content and the data-descriptor bit are unchanged. check-in: 3caf366005 user: dmestas tags: trunk

Closing a protected branch is a protected move; a damaged sync container is an error, not a crash. check-in: 3bcbc1c903 user: dmestas tags: trunk, version-0.3.24

The first nightly soak flipped one bit in a valid container and brought the process down inside the standard inflate: an integer underflow in its end-of-input bit peek that a truncated stream can arrange, reachable from any peer through every sync body. The decoder is now the tree's own copy of the standard one (Zig 0.16.0, MIT) with the comparison reordered so the stream ends, and its origin noted. With the crash gone the same bytes decoded to 58 bytes reported as good: the flip made a block final, and the zlib footer's Adler-32 was read but never checked, by the decoder or by us. The codec now requires a finished stream and verifies the footer against the output. The exact bytes are a fixed test. check-in: 46fcd75fc3 user: dmestas tags: trunk

The push gate judged tags by the branch they named, so sym-B and branch=B were refused under any sign while a closed tag, which names a check-in, was admitted: an account that could not push onto trunk could close its tip, and the owners' next commit was refused against a closed leaf with no cause named. A closed tag aimed at a held check-in now resolves that check-in's branch and is refused when the branch is protected, under any sign, saying close or reopen. A release name on the leaf, a close off the branch, and a close of an unheld target stay admitted; a resolver fault still fails closed. check-in: ff30ba16e8 user: dmestas tags: trunk

The first full nightly soak found seed 22 at 10000 steps: the crash sweep flipped a bit inside a committed config frame, the store's checksum refused to serve it, and the harness reported a config mismatch because only its content-frame path knew that refusal as detection. The config check now routes rot inside the surviving config frame to the repair, and the repair's cut keeps every frame of either kind that ended before the rot, since config frames interleave with content. check-in: 79059346bc user: dmestas tags: trunk

ReleaseSafe keeps every assert as a detector; the build that ships is ReleaseFast, where only the simulation's own state-replay oracle can catch a bug that appears when the checks are gone. The nightly now runs the same DST sweep on both builds. DST_FAST=0 skips the second. check-in: 4df2fad0e6 user: dmestas tags: trunk

The runners keep Zig's cache outside the workspace, so the find under .zig-cache came up empty and xargs handed ls no arguments: ls listed the repo root and the newest entry, src, was run as the harness. The lookup honours ZIG_LOCAL_CACHE_DIR and an empty find is a build failure. check-in: b56ed097fd user: dmestas tags: trunk

scripts/soak.sh passed the optimize flag this Zig rejects, so it has been unrunnable since the toolchain moved; it takes --release=safe now. A scheduled workflow runs it at 03:00 UTC on the x64 runner with the script's nightly budgets, overridable from the dispatch form, keeps the log and any reproduce line as an artifact, and is pinned to the ci toolchain by ci_test like the other workflows. check-in: fa78377389 user: dmestas tags: trunk

The adversarial token generator capped fields at 24 bytes, and the igot parser has required a 40- or 64-byte name since the shun work, so the parser's accept arm was unreachable and the harness's own reachability floor said so at 256 rounds. CI swept 100 rounds, below the floor, and never saw it; the first bounded soak did. One field in four is now hash-shaped, and ci sweeps every target at the floor's round count. check-in: af292c3091 user: dmestas tags: trunk

The transport answers an HTTP Basic challenge; a gated first round is no longer fatal. check-in: 2a3106fcaf user: dmestas tags: trunk, version-0.3.23

A forge that gates a read answers a clone's first request with 401. The login card cannot answer that round, since its secret needs the project code the first reply teaches, and the status was fatal, so the credential never reached the wire. When a verb holds a plaintext password the transport keeps a Basic credential ready, sends nothing until a 401 names the Basic scheme, then repeats the round with the Authorization header and carries it on every later round, as fossil's --httpauth does. A 401 now says what the peer wanted. Proven by an e2e with an authenticating gate in front of zit serve. check-in: 5f541eccc7 user: dmestas tags: trunk

clone and pull authenticate; a private repository is readable. check-in: d75c7a741d user: dmestas tags: trunk, version-0.3.22

Only push and sync signed a login, so a private repository on a forge that gates reads was unreadable by the account that owned it, and a stored login sat unused while pull read anonymously. All four verbs now resolve the login the same way: --user, the URL's user@, then the stored login. A clone holds name and password until the round-0 reply teaches the project code, then signs every later round; a refused round 0 that taught the code is retried with the login, as fossil's client does. The accepted login is stored in the clone. A URL user with no password source is a hint and goes anonymous with a note; --user without a password is an error. Smoke S17 drives a read-gated fossil. check-in: 1d5a6c8fa7 user: dmestas tags: trunk

A multi-check-in fossil push lands; --httptrace reaches a zit server. check-in: 575cd3fcdd user: dmestas tags: trunk, version-0.3.21

fossil --httptrace sends its cards uncompressed under this type, and the server refused it with a bare 415 on the premise that the reference client never sends it. The exchange now tells handleXfer whether the body is a zlib container or plain cards; the reply is framed as before, and fossil reads the reply's framing off the reply's own content type. x-fossil-uncompressed stays reply-only. check-in: cc96464a51 user: dmestas tags: trunk

fossil sends a batch of check-ins as one round with each manifest a delta against the previous one, and the scan reconstructed every pushed delta against the committed store, so the basis was never there and the round was refused as a malformed file card. Every multi-check-in push from a stock fossil failed. The scan now parks a delta whose basis is not held and settles it once the whole round is seen, to a fixpoint, so order within the round does not matter. A basis absent from both the store and the round is refused by name: error delta basis not held: TARGET against BASIS. Smoke S16 pushes three fossil check-ins in one round and checks all three landed. check-in: df556319a8 user: dmestas tags: trunk

The tracker is disposable and does not live with the source, so a number is a dangling pointer. Every zit#NN and #NN reference in code comments, test names, help text, smoke labels, identifiers, release notes and errata is replaced by enough context to stand alone, or by a short plain statement of the problem. No behaviour changed. check-in: 9033c13f21 user: dmestas tags: trunk

The ABNF and §10.1 already make every error card one fossilized TEXT; the unknown-operator prose writes "error bad command: LINE" and reads as four tokens. Records the wire form and the two-token client rule. check-in: ec7d3fc9f7 user: dmestas tags: trunk

Server error cards are encoded so a fossil client reads a refusal; remote refuses a non-URL. check-in: d11082197c user: dmestas tags: trunk

zit remote list saved the word list as the remote and reported success; the loss surfaced at the next push. The help already defines the argument as an http:// or https:// address, and the parser now holds it to that with a usage error that says there are no subcommands. A valid URL is still stored verbatim. check-in: c280025388 user: dmestas tags: trunk

The bad-command and malformed-card refusals were written with raw spaces, so on the wire they were four tokens. A stock fossil client reads an error card as two, reported the rest as unknown commands, dropped the refusal and printed its success summary; a refused push looked landed and re-sent the same artifacts on every autosync. Every error card is now assembled at one point with its text encoded, as fossil's own server writes it. A test drives each refusal through the emitter and fails on a raw space after the operator. check-in: 8f5dff81b0 user: dmestas tags: trunk

log opens once, on the zero-parse image; the open mode is declared, never defaulted. check-in: 61ae22616f user: dmestas tags: trunk, version-0.3.19

log reached past the open funnel and called Repository.open with the default options, which is the full record replay meant for write verbs, and did so twice for a symbolic name. On pkgsrc that was 0.94s before the first row; on the image path it is 0.05-0.10s, output byte-identical on every route. The open mode has no default any more: every direct caller declares it, so the next verb written past the funnel cannot inherit the expensive one silently. An e2e runs every log route under the derive-footprint seam and fails if any of them replays a record. check-in: 0910bc00f5 user: dmestas tags: trunk

A push trigger naming only a tag filter runs for tag pushes alone, so the tags-ignore alone ran nothing for two commits. branches: ['**'] restores every branch; the pin now allows exactly that form. check-in: c004c81e03 user: dmestas tags: trunk

The runners are self-hosted but checkout cleaned the workspace, so every step recompiled the same sources into its own binary from nothing; unit tests and e2e were 6 and 4 minutes on the ARM64 critical path, mostly compile. Zig's cache is content-addressed, so it now lives outside the workspace and warms across runs, in both workflows. The toolchain tarball is kept the same way and still verified against the pinned digest every run. Also pins the tag-ignore trigger the previous commit introduced, which the workflow pin test had rejected. check-in: 3106100d42 user: dmestas tags: trunk

A release tag lands on a commit whose branch push already ran ci, and the tag push ran it again on the same three single-slot runners the release workflow was queueing for. Every branch push still runs; the release workflow, triggered by the tag, gates what it ships. check-in: e9e2ba4431 user: dmestas tags: trunk

The file-history index is built by the derive at ingest, fossil's way. check-in: 09d191e634 user: dmestas tags: trunk, version-0.3.18

fossil fills mlink as each check-in arrives and answers finfo by lookup. zit's derive computed the same row per check-in and dropped it; the index was rebuilt by a separate pass on the first PATH query, 6.8 min on pkgsrc and not resumable. The derive now hands each change set to an index builder as it links, stores it beside the checkpoint, and extends it over the tail on append; a cache hit computes nothing. The on-query build is the fallback and checkpoints every 8192 entries. Index format 4 records completeness so a partial checkpoint is never adopted whole. crosslink reports a tree's files instead of unknown. pkgsrc, fresh clone to first answer: 11.7 min -> 7.1 min + 1s. check-in: df99743ca4 user: dmestas tags: trunk

A push with no URL keeps its credential; the tree-manifest index build is proportional to what changed. check-in: dbfe2adebc user: dmestas tags: trunk, version-0.3.17

Two tree check-ins were diffed by flattening both to full file lists, files-per-check-in per check-in: 34 minutes once on a pkgsrc mirror. The S-card runs are now walked side by side, skipping every sub-manifest whose hash did not move and resolving only where it did. Deletes are held to the end so a moved partition key cannot make a live file look deleted. Trees flatten on demand only for the root and flat/tree pairings; the derive feed never flattens. pkgsrc: 34.3 min -> 6.8 min. check-in: ea4abe48fd user: dmestas tags: trunk

push --user U --password P with no URL parsed to a payload-less no_url, which the verb rebuilt as a bare empty URL, dropping the credential; the push went out anonymous and a peer requiring a login refused it. The parser now mirrors sync: an omitted URL is an empty one, resolved to the saved remote at run time, and the credential comes through. S12 asserts an authenticated no-URL push and a stored-login push both land on fossil. check-in: 4dba18d355 user: dmestas tags: trunk

File index works on submanifest stores; artifact shows S cards; a saved login authenticates a push that names no user. check-in: 8cfa57ecd3 user: dmestas tags: trunk, version-0.3.16

push --user saved the derived secret, but the resolver returned anonymous the instant --user was absent, so autosync (which pushes with no arguments) always went anonymous and a peer requiring a login refused it. With --user absent the saved login now stands in, never prompting; --user "" still forces anonymous. push, sync and config pull share the resolver. Whether such a push lands on a fossil peer is zit#196, a separate pre-existing matter. check-in: e37768d073 user: dmestas tags: trunk

On a submanifest store every check-in is a tree manifest, which the change-set capture refused as unknown, so the index recorded the whole log as unknown and a PATH query reconstructed per check-in anyway -- 39 minutes on a pkgsrc mirror. Trees now flatten inside the diff memo through the Repository seam, so the untouched diff sees a flat list; a tree that cannot be flattened stays unknown. Index format 2 -> 3 so a stale all-unknown blob is rebuilt once. check-in: 001c73defa user: dmestas tags: trunk

A tree check-in carries its file set entirely by S cards and the card renderer had no S lane, so it printed as a manifest with no files while ls on the same hash listed thousands. One lane, in card order, checked against --raw. check-in: da2598d1ca user: dmestas tags: trunk

The v1 index recorded no coverage, so every commit threw it away and re-walked the whole log. v2 records n_rids, the entry count the index covers, and a later query walks only the tail past it. A fresh query walks the new check-ins; a repeat query walks none. check-in: 788b69f097 user: dmestas tags: trunk

The negative-space pin requires every build.zig step to be either invoked by CI or listed with the reason it is not. The cross step was neither, so it failed all three runners -- the pin doing exactly its job on the commit that added an unregistered step.

CI compiles the real thing on all three shipped targets, so cross-compiling them again proves nothing new; the step is for the developer, before the push. Recorded as such. check-in: 770e641725 user: dmestas tags: trunk

This project links no libc and macOS links libc always, so a call that demands it compiles on a mac, passes every local lane, and refuses on both Linux runners -- the discovery order the fmt check already exists to prevent, and what zit#188 spent a red main on.

`zig build cross` compiles the CLI for x86_64-linux and aarch64-linux and runs nothing; gate depends on it. Verified to discriminate: reintroducing std.c.getenv fails cross while a plain macOS build still passes. check-in: 31c4850b95 user: dmestas tags: trunk

std.c.getenv demands libc linkage, which this project does not take, so both Linux runners refused to compile while macOS -- which always links libc -- built it and the local gate passed. cli.Ambient.environ documents exactly this rule; the seam existed and was bypassed.

Read $ZIT_NO_FILE_INDEX in main from init.environ_map like every other seam and hand the runner a flag. Cross-compiles clean to x86_64-linux and aarch64-linux. check-in: 1317aa9864 user: dmestas tags: trunk

The predicate reconstructed a manifest per visited check-in, so one file's history on a mirror of SQLite cost minutes and grew with the history behind it. Persist what the derive already computes and throws away: captureFileFacts is Fossil's mlink row, so invert it by path into .zit-cache/finfo, keyed to the log tail and shun fingerprint like its sibling sidecars.

The PATH walk now traverses the derived adjacency, as the unfiltered walk has since #98, and the index narrows it. The index only NARROWS: an F-card-level diff over-reports a delta that restates a path with the hash it already had (one check-in in 2,113 on SQLite), so every candidate is still confirmed by the B-log-2 predicate and the output is byte-identical. --long reads one manifest per emitted row instead of per visited node.

README.md 72s to 1.6s, btree.c 25s to 1.1s, reconstructions 25,106 to 141. check-in: 17c3b0aaa2 user: dmestas tags: trunk

logChangesPath computed a check-in's effective content for PATH twice — once as `here`, once as the `parent` of the child below it — and each call re-read, re-verified and re-parsed the manifest because fileAt takes neither the walk's plaintext cache nor a baseline memo. One file's history on a mirror of SQLite took 129s.

Thread the walk's cache, a baseline memo, and an answer memo keyed by rid through the predicate, and through timeline -p, which shares the defect. Reconstructions per PATH walk fall from 2N-1 to exactly N; README.md 72s to 24s, btree.c 25s to 9s, timeline -p 16s to 7.5s, output byte-identical. check-in: 184cee5a2e user: dmestas tags: trunk

The write side of the A card. add stores the blob then the attachment artifact in one commit and prints both hashes; rm ends the association with a sourceless card; list folds latest-per-(target, name). The new attachment-size-limit setting (config key 17) caps add, fossil's semantics with a 1 MiB default. A quarry attachment stub lists as external and zit artifact describes it. Smoke S15 pushes an attachment to real fossil and reads it back from fossil's attachment table. check-in: 170a2e88d0 user: dmestas tags: trunk

Once a pull finds 100 or more held artifacts no cluster names, fold them into cluster artifacts — sorted, 800 M cards each, md5 Z — at fossil's create_cluster point and thresholds. They persist with the round and extend the index; the next reply announces the clusters, not the members. A zit-born repository clusters itself the first time anyone pulls from it: 161 artifacts, one cluster, 201 bytes on the next pull. check-in: 10c905d27b user: dmestas tags: trunk

A pull or push announced every held artifact with igot, every round — 1.25M lines, ~63 MB per direction on pkgsrc. Clusters are igot in bulk: index the held clusters (.zit-cache/clusters, keyed to the log prefix, kept warm by persist), announce held minus what they name, seed the promises they made that this repository never collected, absorb pushed clusters server-side so the reply begs for their members, and honour send-catalog (full inventory) and req-clusters (enclosed cluster igots). A no-op pkgsrc pull now sends 4.7 KB. Cluster promises the peer cannot keep are a warning, not a failed pull. No format changed. check-in: e1ca72a8ef user: dmestas tags: trunk

A pull walked every held artifact in persistStore, reading each once from the lazy disk-backed holdings store just to early-return it: 194s of a 197s no-op pull at 1.25M artifacts, reconstructing nothing. Received artifacts are appended beyond the holdings' log-backed prefix, so persist now begins the walk there and processes only the received set. A no-op pull persists nothing; an incremental pull persists what it received, not the millions it already holds. No-op pkgsrc pull 209s to 7s. check-in: a25dcf28f5 user: dmestas tags: trunk

emitReply walked every held rid and, for each, linear-scanned the peer's announced list to suppress a redundant igot — O(held × announced), so a no-op pull where both sides hold everything cost 760 s of server CPU at 1.25M artifacts. Build the announced/requested names into one set per reply and probe it once per rid. ZIT reports the probe count; a test pins it to the held count at two sizes regardless of what the peer announced. check-in: a93f014321 user: dmestas tags: trunk

One bit per log frame in .zit-cache/dispositions: crosslink-relevant or not. A full derive consults it and never reconstructs the file blobs and SUBMANIFEST tree nodes that were most of a wide store's walk; the walk distills a fresh map either way. Same trust tier as the log-scan index: self-checksummed, keyed to a validated log prefix, stamped with the shun fingerprint, absent on any doubt. pkgsrc submanifest cold derive 302 s to 22 s; a consuming derive's body is byte-identical, pinned by test. check-in: a18d2edd67 user: dmestas tags: trunk

repocache.store and checkpoint.store concatenated header + body + image into a full resident copy before writing; both now stream the pieces to the file with the checksum folded incrementally, byte-identical output. The clone path frees its view before the reopen builds a second derived state. ZIT_DERIVE_FOOTPRINT reports the derive body's composition. pkgsrc cold open: peak RSS 2.32 GB to 2.07 GB. check-in: 3107e23e24 user: dmestas tags: trunk

A dense clone no longer holds the repository in memory until the exchange ends: each round's artifacts land in the log and their resident bytes are released, the shape of git's index-pack. A delta whose basis has not arrived defers on its hash; a basis first seen as a phantom is revisited when its content lands. Cone-scoped clones keep the batched pass, whose keep set needs the converged store. Streamed and batched logs are byte-identical, pinned by test. ZIT_PERSIST_WAITING reports the drain's peak deferred set. check-in: 533b0a6490 user: dmestas tags: trunk

Fossil-dialect globs treat ! as a class member; only ^ negates. open reads a versioned-only manifest setting out of the check-in and remembers --manifest=off so update cannot re-enable stamps. add --force adds a named ignore-glob match; directory expansion always honours ignore and clean globs. .fos root names compare case-sensitively on Linux. check-in: b7e077e430 user: dmestas tags: trunk

The crosslink record now carries each changed file's artifact hash (fossil's mlink shape), so annotate reads blobs by hash instead of re-deriving every manifest on the walk. merge3 moves to a linear-space Myers LCS and annotate caches plaintext and baselines across versions. Repocache format 5, checkpoint format 5: the first open after upgrading re-derives. Help states the 12-character prefix annotate actually prints. check-in: a6c2215859 user: dmestas tags: trunk

The crosslink record, the state and the image now carry each check-in's Q cards (sign and target), so the cherrypick/backout line needs no manifest read. Record format 4, image layout 7. info NAME 16 ms to 6 ms on libspatialite, level with leaves; output unchanged. check-in: adb05cc14e user: dmestas tags: trunk

A cache whose image is absent or at an old layout replayed the record body on every fast_read open, since only a miss stored. The hit now serializes the image from the view it just built and stores it, so the fallback is paid once. check-in: 98c5d6c61a user: dmestas tags: trunk

The image has carried adjacency since layout 2 and comment/login since layout 6, so every default row renders from the zero-parse path; only full/-v/-p still read manifests. 60 ms to 8 ms on a 22k check-in store, output byte-identical across formats. check-in: 18ffce5678 user: dmestas tags: trunk

One record scan, as annotate, instead of a manifest per check-in: 20,937 reads to 272 on VERSION, 19s to 0.4s. Each version is reported once under its oldest introducer, matching fossil grep's list byte for byte. check-in: 6be0341149 user: dmestas tags: trunk

A front web server that appends "; charset=utf-8" made the clone reply look compressed, so its plaintext was read as a 1.6 GB container and refused as too large. parseMediaType now ignores parameters and case, as fossil does; serve uses it instead of its own copy. check-in: 8488af309e user: dmestas tags: trunk, version-0.3.6

Glob lists follow Fossil's tokenizer and '#' comment lines; crnl-glob, clean-glob, empty-dirs, encoding-glob, dotfiles and mimetypes work at every tier. Dotfiles are hidden by default and Fossil's checkout-state names are never listed. UTF-16 and invalid UTF-8 classify as Fossil's lookslike.c does; tokenizer rules from glob.c. check-in: 982afa3d8f user: dmestas tags: trunk

Every ancestor cost a manifest read and a full-file line diff — ~36k of each on SQLite where ~400 touched the file. The cache records already say who changed PATH (their diff is against the primary parent, so unlisted means byte-equal there): untouched ancestors inherit the parent's content pointer, the engine copies the parent's origins on pointer-identity, and the LCS trims the common prefix/suffix before its O(n*m) table.

Unfinishable on SQLite before; 0.5-1.2s typical, 57s for a 15k-line heavily-edited file (the multi-hunk quadratic middle remains — a real diff core is the named follow-up). Attribution of a repeated line between equally-valid origins can shift, as any LCS tiebreak may.

zit#172 check-in: 22bdf84fd3 user: dmestas tags: trunk

The default-format timeline inflated a manifest per row — the fetch condition still said 'a.format != .oneline' from when oneline was the default, so zit#170's medium default made every row read one for two strings the image now carries. 66.5s to 0.37s for the full SQLite timeline, output identical.

Repository.manifest counts its reads (ZIT_MANIFEST_READS=1 reports on exit), and an e2e runs ten listings over two history sizes and requires identical counts — a per-row reader fails by construction, no timing, no constants.

zit#172 check-in: 09fa1a7fb3 user: dmestas tags: trunk

zit leaves read and parsed a whole manifest per row — through its delta chain — for the comment's first line and the login. A warm store is image-backed and the image carried neither, so commentOf answered null and every row fell back to that read: 472 leaves, one second, on sqlite.

The image now materializes both: the comment CSR-addressed per rid, the login interned (a repository has a handful of committers). layout_version 6 — an older image is a miss and rebuilds, as the disposable cache intends.

0.3.2. check-in: 71d327a695 user: dmestas tags: trunk, version-0.3.2

The derive built the complete effective file list for every check-in and its parent, then compared them — O(files) per check-in, which on pkgsrc is 100k entries twice over, 305k times. Fossil's add_mlink merges the two deltas' own F cards and seeks the shared baseline; do the same, and walk the lists side by side when the baselines differ instead of materializing them. Keeps recently-read manifests parsed, pinned while a step holds them.

zit#172 check-in: 669593f624 user: dmestas tags: trunk