Thema
#git-alternative
Loots, Blogposts und verwandte Themen rund um diesen Tag. Folge dem Tag, damit passende Updates in deinem Orbit bleiben.
1Gezeigte Loots
0Gezeigte Artikel
6Verlinkte Nachbar-Tags
Anschluss-Themen
Wenn du tiefer einsteigen willst, helfen die benachbarten Tags beim Vergleichen und Querlesen.
Loot
Mehr aus diesem Thema
Text: AI-generated
AI-generated · Automatically published by LinkLoot. graft moves history, blame and merges to the server so CI jobs and agents fetch only what they need — no packfiles, no merge queues. AI-generated: This Loot was created and published automatically by LinkLoot and was not substantively reviewed by a human editor. What it does graft is an open-source implementation of the design described in Anthropic's "Redesigning source control" engineering post. It targets monorepos where most clients are always-online agents or CI jobs rather than humans needing offline-capable clones. The core idea: drop Git's assumption that every clone must be a complete repository. In graft, the server handles history, blame, diffs, merge bases, search, commits, rebases and merges — each exposed as an HTTP API. Many jobs never need a local checkout at all. Commits are stored as changesets in a write-ahead log (files changed relative to first parent, pointing at content-addressed blobs), so creating one costs O(size of change), not O(size of repo). Landing is lock-free: gft land rebases server-side and moves the ref via compare-and-swap on a single row — no global lock, no merge queue contention. Derived data structures (skip-list commit graphs, per-path history, rename detection, access-control indexes) are computed through one generic recursive mechanism, either eagerly or lazily, invisible to clients. Fetching uses identical tiers — node daemon, regional cache, origin server — with immutable objects cached at every level. No packfiles, no negotiation, no repository maintenance anywhere. Working copies are optional and sparse. gft mount provides a writable virtual checkout that fetches files on first read, commits only what changed, and switches revisions instantly. Repository and directory-level access control plus review gates distinguish comment-only and docs changes from logic changes. The project explicitly states it is independent: not Anthropic's Scion, contains no Anthropic code, and is not affiliated with or endorsed by Anthropic. Who it helps Teams running large agent fleets or high-throughput CI against monorepos. If your pipeline spends significant time cloning, fetching packfiles, waiting on merge queues, or negotiating refs, graft's tiered want-only fetch and lock-free landing directly address those bottlenecks. The synthetic benchmark (100k files, 20k commits) reports 82x throughput and 97x less CPU for concurrent lands versus Git on the same hardware, though these are author-run numbers on loopback, not independently verified. Getting started Prebuilt binaries for Linux, macOS and Windows are on the releases page. Alternatively, install via Go: The README links to docs/DESIGN.md, which maps each part of the original design post to the implementation. Benchmarks and methodology are in bench/RESULTS.md. Limits and costs graft is Apache-2.0 licensed; the license permits use, modification and distribution under its terms. Hosting infrastructure is separate from the software license — running your own servers incurs whatever compute and storage costs apply. The project is early-stage and self-described as an implementation of a publicly described design; production readiness depends on your operational requirements. No pricing model exists because there is no hosted service. Source links Official repository Project documentation Project license
- LinkLoot access
- Free
- Provider costs
- Unknown
Review open
Blog
Verwandte Artikel
Noch kein Blogpost zu #git-alternative
Aktuell gibt es keinen veröffentlichten Artikel mit diesem Tag. Schau im Blog nach verwandten Themen oder folge dem Tag für spätere Updates.