Bun's Roadmap

[Jarred-Sumner]

Bun's Roadmap

The Bun team updates this issue to share our current roadmap, priorities, and goals. You can see our current update below, and previous updates at the bottom.

Spring 2025

Node.js compatibility

For every change we make to Bun, we run the Node.js test suite to ensure that Bun is compatible with Node.js. We also started a bounty program within our team, to accelerate our progress.

Our goal is to be 90% compatible with Node.js. We'll be providing updates as our progress nears completion.

Features

While we continue to work on Node.js compatibility and bug fixes, we’re also going to be working on new features in upcoming releases of Bun, including:

• #18812 Redis client
• #19699 MySQL support with Bun.sql
• #19701 SQLite support with Bun.sql (in addition to the existing bun:sqlite API)
• #4824 Test runner support in VSCode
• #7589 Running scripts with filters, and in parallel, using bun run
• #947 REPL support (replacement for existing, 3rd-party package: bun repl)
• Improve usage of bun install at work, including:
  • #7157 Support migration from pnpm-lock.yaml to bun.lock
  • #4844 Support catalogs
  • #6608 Support nested resolutions and overrides
  • Fix issues with workspaces, private registries, and corporate proxies

Reliability

We're also hard at work improving Bun's reliability, for example, reducing crashes and memory leaks. We've been working on a few projects to help with this.

• #19057 Address sanitizer (aka. ASAN)

  We now run address sanitizer in our CI, to detect bugs like heap buffer overflows

• v1.2.2 10-30% less memory usage when idle

  We made changes to how Bun schedules garbage collection, so that it syncs with JavaScriptCore. This improves memory usage when Bun is idle.

• Distribute a debug/assertion build of Bun

  Sometimes it's difficult to reproduce crashes without having access to your source code. We will soon distribute a special, optional build of Bun that includes more assertions and debug logs, so that it's easier for you to provide us with reproductions of issues.

Hiring

Bun’s team is growing!

• 3 years ago, it was just Jarred.
• 2 years ago, we raised a Seed round and grew to a team of 4.
• Today, we're 14 people

We’re also announcing new job openings at our San Francisco office, including:

• Head of Propaganda (aka. Developer Advocate) for $120k-$180k + Equity
• Engineering Manager for $180k-$240k + Equity
• Senior Systems Engineer(s) for $160k-$220k + Equity

If you’re excited about working at Bun, you can apply at bun.sh/careers.

Something new

Stay tuned.

Previous updates

Winter 2024

bun install

• Implement a text-based lockfile format #11863 (will unblock add support for bun dependabot/dependabot-core#6528)
• Support nested "resolutions" / "overrides" #6608
• Fixes for frequently-reported issues for things like private git URLs and some edgecases with hoisting
• bun update --interactive (Interactive update packages command. #4895)
• CA support for SSL connection to NPM registries #271
• Support an "engines" field in package.json #5846
• Update bun init to add "engines": { "bun": ... by default. If bun present then ignore node shebangs by default (#9346)
• bun why (like yarn why) #692

Runtime

Node.js compatibility

• Implement node:http2 server to unblock grpc2
• Implement more of V8 C++ APIs to unblock canvas, node-pty
• Rewrite node:http (Rewrite node:http to not use Request & Response #14384)
• Add much more comprehensive test coverage for napi
• Get at least 75% of Node's test suite running on every commit (as of last update: 37% currently). This number will go up in future
• Investigate removing our undici override
• Rewrite node:crypto #13681
• Add CompressionStream #1723

Reliability

• tracking: JSValue safety in our Zig and JS bindings #15141

Bake (Bundler)

The goal for Bake is to make Bun the most productive tool for building static & full-stack JavaScript and TypeScript applications, leveraging runtime, bundler, and transpiler integration to make things simpler.

High level:

• Make a fast HMR development full-stack server - bake dev server tracking issue #14324
• Production builds (bake production build tracking issue #14763)
  • Static production builds
• Implement an integration with a popular framework (such as Next.js)
• Easy & powerful plugin API

bun test

• Reporter API or socket API

Organizational

• Hire a contractor to help us with CI
• Hire a technical writer to help with docs
• Hire an engineering-focused role for maintaining the TypeScript types & frontend for docs + help with integrations with third-party packages
• Hire more systems engineers

Fall 2024

bun install

Essentially, feature complete.

• bun outdated
• Implement bun pm pack #487
• bun publish
• Text-based lockfile format (which will help unblock add support for bun dependabot/dependabot-core#6528)
• bun update --interactive (Interactive update packages command. #4895)
• Fixes for frequently-reported issues for things like private git URLs and some edgecases with hoisting

Runtime

Node.js compatibility

• TextEncoderStream & TextDecoderStream
• V8 C++ API (in-progress)
• Implement node:cluster
• Implement node:http2 server to unblock grpc2
• Rewrite node:zlib to address performance issues in some common packages
• Rewrite node:crypto #13681
• Fix various bugs in node:http, avoid wrapping Bun.serve() and fetch() and use a more direct implementation
• Add much more comprehensive test coverage for napi
• Get at least 25% of Node's test suite running on every commit (as of last update: 15% currently). This number will go up in future quarters.

Reliability

• Delete almost all code that manually reads JSValue. Replace with an IDL bindings generator for JavaScriptCore objects/classes that supports C++ & Zig output, and importantly: function arguments, return values, and exceptions. It should make the lifetime of these values brainless for us to reason about.
• Address TLS-related issues that have cropped up
• Continue to allocate significant time to fixing bugs and improve test coverage as they crop up

Bundler

• Fix many common bugs people run into related to source maps or incorrect output
• A new, higher-level iteration of Bun's bundler designed for server-driven JavaScript bake dev server tracking issue #14324
• CSS parser and bundler MVP of CSS parser #14167

bun test

• Reporter API or socket API

Organizational

• Hire a contractor to help us with CI
• Hire a technical writer to help with docs
• Hire an engineering-focused role for maintaining the TypeScript types & frontend for docs + help with integrations with third-party packages
• Hire more systems engineers

[RnbWd]

this seems like a lot of work... wow

[space7panda]

Can't wait for full npm-like config support. We are using our own npm registry, and the current build time is 40-60 minutes. If it will be at least 10 minutes this will be a game-changer!

[officialrajdeepsingh]

I am waiting for the stable version of the bun. What change bun in the javascript world.

[tipiirai]

Also waiting for the stable version. So exciting! I bet it's going to take a lot of time though. The scope of this project is much larger than what Node attempts to solve.

[matepaiva]

can someone confirm that right now I can't use the package mysql2? It throws me an error because it depends on TLS package and on NET package, and I think they were not implemented yet in Bun, but I am not sure.

[Kapsonfire-DE]

still missing websocket server support

[GCSBOSS]

Can a tinitiny docker image be placed somewhere in there? 👀

[MystPi]

Can't wait for Windows support... I'll be anxiously awaiting it until it's out. 😅

[0x4a61636f62]

still missing websocket server support

Dito

[sodeprecated]

Are you planning to support http2?

[moderation]

Are you planning to support http2?

HTTP/3 is on #159 so I would say HTTP/2 will likely be added before that

[sodeprecated]

Are you planning to support http2?

HTTP/3 is on #159 so I would say HTTP/2 will likely be added before that

I would say that http3 on roadmap is more about quic protocol and there is no straight inheritance between http3 and http2, so maybe support of http2 needs to be marked explicitly.

[frodoe7]

Motivated to see a stable version ASAP

[LaurelineP]

Promising !
Any roadmap on a progressive translation ?

[Parvat-R]

When is the support for windows coming? Cant wait for it.

[jacoscaz]

I would say that http3 on roadmap is more about quic protocol and there is no straight inheritance between http3 and http2, so maybe support of http2 needs to be marked explicitly.

Another vote for HTTP/2 as it's a prerequisite to gRPC support.

[dyaskur]

Any plans to create a desktop app like electron?

[xhyrom]

When is the support for windows coming? Cant wait for it.

#43 you can use WSL for now

[kravetsone]

Why not bun update -i not by default? I guess it would be more convinient

--interactive can be short alaised to -i I am looking to do this but I need some help from @Jarred-Sumner or someone to get to me a basic idea what it will look like and etc.
#4895 (comment)

I mean interactive by default not in CI without flagging

Not really because interactive should be optional that’s how pnpm and yarn do it is where they update everything and if you need interactive then you use the flag.

but what usecase to not use interactivity? i guess it more better to know what u bumps

also just add --all flag to bump all deps without interactivity

[Eveeifyeve]

Why not bun update -i not by default? I guess it would be more convinient

--interactive can be short alaised to -i I am looking to do this but I need some help from @Jarred-Sumner or someone to get to me a basic idea what it will look like and etc.
#4895 (comment)

I mean interactive by default not in CI without flagging

Not really because interactive should be optional that’s how pnpm and yarn do it is where they update everything and if you need interactive then you use the flag.

but what usecase to not use interactivity? i guess it more better to know what u bumps

also just add --all flag to bump all deps without interactivity

Because interactively is bearly used and will be a pain in way when you just trying to update all deps. By default it should be updates all deps but you have the option to interactively update deps using -i or --interactive

[danielniccoli]

@kravetsone It is idiomatic in Linux and Windows that shell commands should be non-interactive first if they are meant to be used in both ways. And for bun this is certainly the case.

@Eveeifyeve While it is good to have your opinion formed by outside factors, I'd certainly say that there are better arguments than looking at yarn or pnpm to decide what bun should do (see my above argument). Whether or not interactive is bearly used needs some sources and whether it is a pain is certainly subjective. :)

I would recommend to move that discussion into the Discussions section. Saying that, commenting on the matter makes me guilty as well :)

[Eveeifyeve]

@Eveeifyeve While it is good to have your opinion formed by outside factors, I'd certainly say that there are better arguments than looking at > yarn or pnpm to decide what bun should do (see my above argument).

I will look at your above argument.

Whether or not interactive is bearly used needs some sources and > whether it is a pain is certainly subjective. :)

I can't find some sources, but I know there is some data out there.

I would recommend to move that discussion into the Discussions section. Saying that, commenting on the matter makes me guilty as well :)

Or the bun discord using the threads thing.

[RobinKnipe]

Why not bun update -i not by default? I guess it would be more convinient

If you want to change the way a command works for you, why not just create an alias? e.g.

# in ~/.bash_aliases
alias buni='bun update -i'

[Eveeifyeve]

@danielniccoli You ever get an answer on your streams inquiry? You are talking about WHATWG Streams, correct?

Um This is not related idk why you even mentioned this.

@Jarred-Sumner @paperdave.

[Eveeifyeve]

Um This is not related idk why you even mentioned this.

@Eveeifyeve

Clearly for clarification.

WHATWG Streams are what's going on circa 2024, not necessarily Node.js-specific "node:stream".

Just report it if it's an issue if it's a discussion please take it to discussions on GitHub (which I don't know is enabled) or bun discord.

[vithalreddy]

Hi @Jarred-Sumner

Congrats on Bun 1.2 release, Appreciate Bun Team's efforts.

Could please take some time to update on the Bun's roadmap, what's been done and what's in the store for bun in near future?

[sourabratabose]

Can we please get a bun built in linter and formatter ? Deno does it, they just adopted dprint and biome I guess. Those 2 are very basic features and at this point of developement of JS apps, pretty much needed everywhere. Looking forward to bun linter and formatter.

[RedStar071]

Can we please get a bun built in linter and formatter ? Deno does it, they just adopted dprint and biome I guess. Those 2 are very basic features and at this point of developement of JS apps, pretty much needed everywhere. Looking forward to bun linter and formatter.

It would be a killer feature like that, it would be great

[graffhyrum]

Maybe by leveraging this?
https://x.com/colinhacks/status/1883907825384190418
https://standardschema.dev/

[Akalanka47000]

I'm not sure if I'm asking for much but will it be possible by any chance to prioritise addressing memory leaks in Bun as of its current version. It would give a lot of us the possibility to use Bun in production without any worries since it's feature rich enough for most use cases at this point. Right now memory usage is quite unstable from what I could notice, for example when compared with 1.2.1, 1.2.2 is leaking more memory in one of my codes, and the leak of 1.2.1 was there since 1.1.15. I noticed similar fluctuations multiple times in between version upgrades.

[Spixmaster]

Can we please get a bun built in linter and formatter ? Deno does it, they just adopted dprint and biome I guess. Those 2 are very basic features and at this point of developement of JS apps, pretty much needed everywhere. Looking forward to bun linter and formatter.

Just use the already existing formatters and linters. Why bloat Bun?

[Eveeifyeve]

Can we please get a bun built in linter and formatter ? Deno does it, they just adopted dprint and biome I guess. Those 2 are very basic features and at this point of developement of JS apps, pretty much needed everywhere. Looking forward to bun linter and formatter.

Just use the already existing formatters and linters. Why bloat Bun?

Well bun is already bloated with the SQL drivers and features, so I don't get your point that your trying to prove, I think they need to focus on NodeJS compatibility and Memory Leaks before making any new feature, it's nice and all but I feel like bun could be used in way more environments if there isn't memory leaks everywhere and NodeJS compatibility was fully done.

[Spixmaster]

Can we please get a bun built in linter and formatter ? Deno does it, they just adopted dprint and biome I guess. Those 2 are very basic features and at this point of developement of JS apps, pretty much needed everywhere. Looking forward to bun linter and formatter.

Just use the already existing formatters and linters. Why bloat Bun?

Well bun is already bloated with the SQL drivers and features, so I don't get your point that your trying to prove, I think they need to focus on NodeJS compatibility and Memory Leaks before making any new feature, it's nice and all but I feel like bun could be used in way more environments if there isn't memory leaks everywhere and NodeJS compatibility was fully done.

Your priorities are reasonable. However, the argument that Bun is already bloated and hence advocate for even more bloating does not resonate with me. Extending a standard library has nothing to do with with integrating external applications with a completely different purpose.

[kravetsone]

Can we please get a bun built in linter and formatter ? Deno does it, they just adopted dprint and biome I guess. Those 2 are very basic features and at this point of developement of JS apps, pretty much needed everywhere. Looking forward to bun linter and formatter.

Just use the already existing formatters and linters. Why bloat Bun?

Well bun is already bloated with the SQL drivers and features, so I don't get your point that your trying to prove, I think they need to focus on NodeJS compatibility and Memory Leaks before making any new feature, it's nice and all but I feel like bun could be used in way more environments if there isn't memory leaks everywhere and NodeJS compatibility was fully done.

Your priorities are reasonable. However, the argument that Bun is already bloated and hence advocate for even more bloating does not resonate with me. Extending a standard library has nothing to do with with integrating external applications with a completely different purpose.

the idea is interesting
its also helpful in DX

but for now yeah bun should more focus on stable

[kravetsone]

Can we please get a bun built in linter and formatter ? Deno does it, they just adopted dprint and biome I guess. Those 2 are very basic features and at this point of developement of JS apps, pretty much needed everywhere. Looking forward to bun linter and formatter.

Just use the already existing formatters and linters. Why bloat Bun?

Well bun is already bloated with the SQL drivers and features, so I don't get your point that your trying to prove, I think they need to focus on NodeJS compatibility and Memory Leaks before making any new feature, it's nice and all but I feel like bun could be used in way more environments if there isn't memory leaks everywhere and NodeJS compatibility was fully done.

did u see memory leaks Bun@>=1.2.2?

after this version i see more great memory stats

did u use napi (prisma for example)?

[gland2015]

Great job，I think the most important task now is to complete the bake (bundler)

[Christophvh]

@Jarred-Sumner Any possibility to update this? Would love to see future roadmap.

[kravetsone]

Now it is updated. Great!
But also I will like to see here test coverage percent

[Eveeifyeve]

Might also want to look into supporting the wrangler cli because that is used for stuff like R1 and Cloudflare Pages which will require workers to be stablized.