Welcome to the Buf Schema Registry (BSR)! This overview covers what the BSR is and the challenges it aims to solve within the Protobuf ecosystem. If you just want to see the BSR in action with code examples, check out The tour.
The BSR stores and manages Protobuf files as versioned modules, so that individuals and organizations can consume and publish their APIs easier than ever before.
The BSR comes with a browsable UI, dependency management, API validation and versioning, autogenerated documentation, and an extensible plugin system which enables remote code generation.
Centralized Registry - The BSR is the source of truth for tracking and evolving your Protobuf APIs. A centralized registry enables you to maintain compatibility and manage dependencies, while making it easy for your clients to consume APIs reliably and efficiently. Having a centralized, Protobuf-aware, registry has the added benefit of protecting against broken builds.
Dependency Management - The BSR finally introduces dependency management to the Protobuf ecosystem. You can now declare, resolve and use hosted BSR modules as dependencies in your projects.
Put simply, you don't need to copy your
.protofile dependencies around anymore. The
bufCLI interacts directly with the BSR to fetch your dependencies, (analogous to
cargofor Rust, and Go modules in Go).
UI and Documentation - The BSR offers complete documentation for your Protobuf files through a browsable UI with syntax highlighting, definitions, and references.
Code Generation - The BSR makes possible two distinct forms of code generation:
Local Code Generation: The BSR contains remote references that allow you to immediately generate code for your language of choice. Use your existing
buf generate, and you're ready to go.
Remote Code Generation (experimental): Don't want to manage plugins? No problem. The BSR offers remote code generation through hosted plugins and templates. The remote autogenerated code gets stored in a managed software repository that you fetch like any other library with tools you already know:
npm install, etc.
You're probably wondering why adopting the Buf Schema Registry (BSR) is an improvement over your existing Protobuf workflows. Above we highlighted what features are available, but let's break down why the BSR aims to solve existing problems.
The Protobuf ecosystem deserves build guarantees
Traditional workflows push Protobuf files to version control systems, but these systems lack Protobuf-awareness and thus consumers often waste time working with Protobuf files that don't compile. Yes, some organizations add checks to catch broken Protobuf files, but these are error-prone and don't scale well because each repository needs to be configured, setup and maintained.
Since the BSR is a Protobuf-aware registry, it prevents Protobuf files that don't compile from being pushed to the origin in the first place. Your consumers will have confidence that Protobuf files consumed from the BSR are not broken and can compile. Everyone in the ecosystem benefits because compilation guarantees are pushed from the individual to the BSR.
Generated documentation for all
Sadly, easy to consume and up-to-date documentation is rarely available for Protobuf files. Some organizations setup workflows to generate documentation, but this is yet another manual step that has to be configured, setup and maintained. Furthermore, much of the tooling and plugins are unsupported and generate incomplete documentation.
The BSR comes built-in with autogenerated documentation. You get live documentation for every commit to the BSR. Which means live and up-to-date documentation for latest and historic commits. Even better, the documentation the BSR provides has syntax highlighting, definitions, and references.
Keeping Protobuf files synced
Every organization that adopts Protobufs needs to solve distribution, whether internally across teams or externally exposed to the public. Protobuf files are usually checked into repositories, often dispersed, and it becomes challenging to keep Protobufs synced across projects. API drift is a common issue and even worse, forked Protobuf repositories accidentally get consumed by downstream dependents (instead of the upstream). It's a mess.
The BSR solves this by offering a centralized registry to store all your Protobuf files, simplifying the process of publishing and consuming. By making the BSR the single source of truth it is possible to power developer workflows and business processes around Protobuf without worrying how to keep everything in-sync.
Client SDKs should not be an afterthought
Define. Generate. Consume.
Defining a Protobuf-based API enforces a contract between producer and consumers. The issue is that consumers are typically an afterthought in the process.
Before a client can consume a Protobuf-based API they need to generate an SDK for their language of choice. Traditionally consumers are left to figure out how to build and generate clients, but this is often cumbersome as little guidance is provided and furthermore not all Protobuf files correctly encode options for a given language.
The Buf ecosystem solves this by decoupling consumer and producer concerns. The
buf CLI can modify language-specific options on the fly, and the BSR facilitates generating client SDKs by offering local and remote code generation.
Generating an SDK client should be as simple as keeping code up-to-date like any other library in your language of choice.
Once you've installed the latest version
buf, you're ready to use the BSR!
Here are a few ways to get moving:
Dive directly into the docs to get familiar with the BSR and the various components.
Learn how to run
bufcommands that leverage the BSR.
The tour will give you an overview of the BSR, and takes approximately 20 minutes to complete.
Remote Generation is currently an experimental feature. This feature lets the BSR generate data types and service stubs for your modules and store the generated code in a hosted registry. With this, you can consume the generated code like any other library in your language of choice.
Note, this feature is currently available for Go. But we have plans to add additional support for other languages.