---

ByteORM is a lightweight ORM for Rust that generates a fully typed client crate from a Prisma-like .bo schema. Define your schema, run byteorm push, and use a typed Rust client with query builders, CRUD operations, and connection pooling.

Using ByteORM

1. Initialize

byteorm init

This creates byteorm.toml, a starter schema.bo, and a generated-client ignore entry. Existing projects that already use byteorm/*.bo or generated/ keep that layout.

Example byteorm.toml for a single schema:

[schema]
path = "schema.bo"

[client]
output = ".byteorm/client"
crate_name = "byteorm-client"
dependency_source = "vendored"

Example for multi-schema projects:

[schema]
directory = "byteorm"

[client]
output = "generated"
crate_name = "byteorm-client"
dependency_source = "vendored"

ByteORM v0.1.x keeps dependency_source = "vendored" as the default. The generated client includes its matching byteorm-macros copy, so projects do not need separate crates.io packages yet.

2. Define Your Schema

Create or edit schema.bo:

enum PostStatus {
    Draft
    Published
    Archived
}

model User {
    id         BigInt      PrimaryKey
    email      String      Unique
    username   String      NotNull
    created_at TimestamptZ @default(now())
}

model Post {
    id         BigInt      PrimaryKey
    user_id    BigInt      NotNull ForeignKey(User.id, onDelete: cascade)
    title      String      NotNull
    content    String      NotNull
    status     PostStatus  @default(Draft)
    created_at TimestamptZ @default(now()) Index
    updated_at TimestamptZ @default(now())
}

3. Generate Or Push

Generate the typed client without touching the database:

byteorm generate

Push schema changes to the database and regenerate the client:

byteorm push

If a migration would drop tables, columns, or enum types, ByteORM stops unless you explicitly accept it:

byteorm push --accept-data-loss

Inspect resolved paths and config:

byteorm doctor

Generate shell autocomplete scripts:

byteorm completions
byteorm completions install
byteorm completions powershell
byteorm completions zsh
byteorm completions bash

byteorm completions detects the current shell and prints the recommended install command. byteorm completions install writes the completion script to the standard location for bash, zsh, fish, PowerShell, or elvish.

For a one-off bash session without installing anything:

source <(byteorm completions bash)

The install scripts also detect the current shell and ask whether to install autocomplete automatically. Set BYTEORM_INSTALL_COMPLETIONS=1 to install without prompting, or BYTEORM_INSTALL_COMPLETIONS=0 to skip it.

4. Use The Client

Add the generated client path printed by byteorm init or byteorm push to your Cargo.toml:

[dependencies]
byteorm-client = { path = ".byteorm/client" }

Then use it from Rust:

use byteorm_client::{Client, PostStatus};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
    let client = Client::new("postgres://user:pass@localhost:5432/mydb").await?;

    let user = client.user.create(|u| {
        u.set_email("[email protected]")
            .set_username("alice")
    }).await?;

    let post = client.post.create(|p| {
        p.set_user_id(user.id)
            .set_title("Hello ByteORM")
            .set_content("This is my first post using ByteORM!")
            .set_status(PostStatus::Draft)
    }).await?;

    println!("Created post: {} (id={})", post.title, post.id);

    Ok(())
}

A complete working example lives in examples/blog/. Run it with cargo run --example blog from that directory.

Editor Support

The IntelliJ MVP lives in integrations/intellij/byteorm-intellij/. It is intentionally kept as a normal subproject rather than a git submodule so the editor-support code stays close to ByteORM while remaining easy to extract later into a dedicated repo for IntelliJ, VS Code, Zed, or other editors.