Artifact fe8cb7de25690cfad5805dc7a621f513af8e56a5c33e5c0d835dff77387263fd:
- File
docs/site/content/docs/guides/migrating-from-fossil.md
— part of check-in
[f747958447]
at
2026-10-29 00:10:09
on branch feat/attachments
— feat(attachments): add, delete, list and read attachments as fossil does
AttachmentAdd, AttachmentDelete, Attachments and AttachmentRead write and read the artifacts fossil 2.28's attach.c writes: the file as one ordinary blob and an A C D U Z control artifact naming it, a deletion as an A card with no source. Targets are a wiki page's exact name or a ticket's or tech note's id or unique prefix; a name two kinds answer to is refused.
AttachmentAddExternal stores Quarry RFC 0016's stub in place of the bytes, with the mimetype on the N card; list and read report the reference and never hand the stub's text back as the file. (user: [email protected], size: 9343) [annotate] [blame] [check-ins using] [more...]
title: Migrating from fossil weight: 20
aliases: ["/docs/migration-from-fossil/"]
This guide is for users of upstream Fossil SCM (the C
binary distributed by fossil-scm.org) who are evaluating libfossil — a
pure-Go implementation of the Fossil repository format and sync protocol.
It is intentionally honest about gaps. If a capability is not listed under "Command coverage" below, assume it is not implemented yet.
For a structural overview of how libfossil is organised — the package layout, storage model, and pluggable axes — see Architecture.
Repo format compatibility
libfossil reads and writes the same on-disk .fossil SQLite repository
format as upstream Fossil. A repository created by either tool can be opened
by the other at the file level — there is no import/export step and no data
migration is needed. Both implementations share the same blob table layout,
manifest/control artifact encoding, and delta chain representation.
One caveat: features introduced in very recent upstream Fossil releases
(bleeding-edge artifact card types, new Fossil-specific SQL functions used in
triggers) may not yet be parsed by libfossil. The tests that compare
against stock fossil run against the pinned release, 2.28. See
Conformance and Testing.
Binary name and invocation
The compiled binary is named libfossil (not fossil) to avoid $PATH
collisions when both tools are installed on the same machine:
go install github.com/danmestas/go-libfossil/cmd/libfossil@latest
libfossil uses a subcommand-grouped syntax: every command lives under the repo group,
and individual operations live beneath it. Upstream Fossil uses a flat
command style. Examples:
| Upstream Fossil | libfossil |
|---|---|
fossil clone URL repo.fossil |
libfossil repo clone URL repo.fossil |
fossil ci -m "msg" file.txt |
libfossil -R repo.fossil repo ci -m "msg" file.txt |
fossil timeline |
libfossil -R repo.fossil repo timeline |
The -R <path> flag selects the repository and is a global flag (before the
repo subcommand group).
Command coverage
The following repo subcommands are implemented today. Every entry below
corresponds to a registered Kong command in cli/repo.go in the libfossil
source tree. libfossil --help lists them all.
Repository creation
repo new— create a new, empty repository file.repo clone— clone a remote Fossil server over HTTP(S). With--coneit clones a sparse store; with--structuralit derives branch and tag state from the origin's horizon seed;--seed-keypins the key the seed must be signed by. See Sparse cones and Horizons.
Working tree and checkout
repo open— materialize a checkout in a directory (creates.fslckout).repo co— check out a named version.repo ls— list files in a version.repo status— show pending changes in the working directory.repo add— stage files for addition.repo rm— stage files for removal.repo rename— rename a tracked file.repo revert— undo staging changes.repo ci— commit the checkout's changes, or only the named files, with any pending merge as a merge parent.--nosignstores one check-in unsigned; see Signing.repo cone— show or edit the checkout's sparse cone (show,add,remove,set,dense,test) and the repository's store cone (store,deepen,compact,cost).
History and inspection
repo timeline— repository history (-nsets the count, default 20).repo log --path— the check-ins that changed a path (fossil finfo).repo snapshot— commit one file's bytes onto trunk without a checkout; see Embedding the CLI.repo diff— show changes vs. a version.repo annotate/repo blame— annotate file lines with commit history.repo cat— output artifact content.repo hash— hash files (SHA1 or SHA3-256).repo info— repository statistics.repo resolve— resolve a symbolic name to a UUID.repo schema— operate on the synced table schema.repo query— run ad-hoc SQL against the repository DB.
Branching and merging
repo branch—ls,newandclose.repo tag—lsandadd.repo merge— merge a divergent version into the current checkout.repo conflicts—lslists files holding merge conflict markers and conflict-fork entries;show,pick,mergeandextractinspect and resolve one conflicted file.repo mark-resolved— resolve a conflict-fork entry. A merge conflict needs no marking: as in fossil, editing the markers out resolves it.repo bisect— binary-search commits for bug introductions (good,bad,next,skip,reset,ls,status).repo undo/repo redo— undo/redo the last operation.repo stash—save,pop,apply,ls,dropandclear.
Collaboration and admin
repo user— user management.repo invite— generate an invite token for a user.repo config— get/set/list repository configuration.repo wiki— wiki page operations.repo uv— unversioned file operations.repo extract— extract files from a version.repo delta— create/apply delta operations.repo verify— verify repository integrity.repo serve— serve the repository over HTTP for clone and sync (-P host:port, default127.0.0.1:8080). Access follows the user table. It is a sync endpoint, not a web UI.repo submanifests—enable,retune,disableandstatusfor sub-manifest trees; see Sub-manifests.
Known gaps vs. upstream Fossil CLI
The following upstream commands have no libfossil equivalent today:
fossil ui/fossil http— no web UI or HTTP admin server is shipped.repo serveanswers sync and clone requests only; it is the counterpart offossil serverfor the sync protocol and nothing else.fossil sync/fossil push/fossil pull— there is no CLI verb; sync is exposed via the Go API (Repo.Sync,Repo.Pull,Repo.XferHandler) but not as a subcommand.fossil update— merging a newer trunk into a checkout has no dedicated CLI command; userepo coplusrepo mergeas needed.fossil search— no CLI front-end for full-text search.fossil fts-config/ FTS index management — no CLI command.
What's not supported yet
Beyond the gaps listed above, several broader features of upstream Fossil are
not yet exposed by libfossil:
- Web UI / HTTP admin — no equivalent of
fossil ui. The sync protocol handler exists asRepo.XferHandler()andrepo serve, but the HTML UI, wiki viewer, ticket tracker, and forum are not implemented. - Interactive conflict resolution — there is no interactive merge
tool.
repo conflictslists, shows and resolves conflicts by picking a version or re-merging; otherwise edit out fossil's conflict markers in your editor. - Full-text search CLI —
internal/searchcontains indexing and query primitives, but no CLI command surfaces them. Use the Go API directly. - Ticket and forum subsystems — no commands or pages. Ticket and forum
artifacts are synced and stored like any other artifact. Tickets have a Go
API:
Repo.TicketAddandRepo.TicketSetwrite the change artifactfossil ticket addandfossil ticket setwrite, andRepo.TicketListandRepo.TicketChangesread tickets back. Report formats, TH1 ticket scripts andprivate_fields are not supported. Attachments to tickets, wiki pages and tech notes have a Go API too:Repo.AttachmentAdd,Repo.AttachmentDelete,Repo.AttachmentsandRepo.AttachmentReadwrite and read the artifactsfossil attachment addand the attachment pages do. - Mirror / export bridges to Git, svn, and other systems — not implemented.
If any of these are load-bearing for your workflow, stay on upstream Fossil for now.
Interop patterns
Where libfossil shines is in environments where the upstream C binary is
inconvenient to run:
- CGo-free CI pipelines —
libfossillinks statically with a pure-Go SQLite driver (db/driver/modernc), so clone/sync steps work on stripped container images and cross-compiled targets with no system dependencies. - Embedding in Go services — the
Repohandle can be used directly from a Go program to read history, extract artifacts, or accept pushes over a custom transport (WebSocket, NATS, libp2p). - WASI / browser deployments — the
db/driver/ncrucesSQLite driver compiles to WebAssembly, enabling repo access from edge runtimes.
See Examples for runnable starting points. The
cmd/libfossil package is also a short working example: main.go wires the
CLI, while driver_modernc.go registers the default driver and
driver_ncruces.go shows the tagged WebAssembly-capable alternative.