Git-Xet is a plugin for git that allows git repositories to scale to terabytes of data.
Content -- both binary and larger text-based files -- are deduplicated across commits, files, and blocks of identical content in otherwise different files. This allows git to efficiently work with repositories containing large datasets that evolve over time. See our blog post here for more information on the technical details of this process, including a link to our CIDR 2023 paper.
Once a repo is set up to use Git-Xet, nothing more is required. No per-file configuration or tracking is necessary.
Using the git xet mount feature, any branch or commit can be mounted locally as a read-only folder. The data in this folder is seamlessly materialized as needed, allowing large repositories to be viewed quickly. This is built using our parallel project, nfsserve.
Git-Xet naturally integrates seamlessly with XetHub, which provides a number of nice perks, such as data analytics and integration with tools like xetcache and pyxet for easy access to your data from any environment. It also integrates with GitHub, where repositories can be quickly enabled to use Git-Xet with the XetData app. Or, just used Git-Xet locally against a local, private data store.
Finally, Git-Xet is an open source tool that is free to use and build upon. By default, the binary data content is stored on XetHub and backed by S3. It can also be configured against a local data store for a fully free and open source route to managing large repositories.
Documentation for Git-Xet, PyXet, and the Xet CLI is available on XetHub here.
The easiest way to install Git-Xet is to install one of the prebuilt binaries. Once the git-xet executable is in your path, run git-xet install from the command line to configure the git config settings, then any enabled repository will work.
To build Git-Xet from source, install the rust toolchain, then run cargo build --release in the rust/ subfolder. The git-xet binary is then in rust/target/release/. After this, run git-xet install to set the proper git config settings.
We will develop this package in public under the BSD license and welcome contributions.
Join our Slack to get involved. To stay informed about updates, star this repo and sign up at XetHub to get the newsletter.
The use of the fully managed XetHub service provides many perks, including reliable data storage and seamless minimal-configuration integration with git.
Go to XetHub to create an account. On your first sign-in, you should get a personal access token that can be used for authentication, or go to https://xethub.com/user/settings/pat to create a new token.
There are two ways to authenticate with XetHub:
Run the command given when you create your personal access token:
git xet login -e <email> -u <username> -p <personal_access_token>git xet login will write authentication information to ~/.xetconfig
Environment variables may be sometimes more convenient:
export XET_USER_EMAIL = <email>
export XET_USER_NAME = <username>
export XET_USER_TOKEN = <personal_access_token>
