I am a big fan of Vite. But I have zero clue what those other tools are. I swear to God, I just put my head down to do some work and all the sudden, frontend tooling has evolved. I wonder if there is a push towards a "boring but works" stack.
the other tools are for testing, bundling, linting and formatting. Previously you would use different tools from very different open source projects for these things, with different configurations, update cycles etc. Now it's all covered by one simple toolchain. Vite+ is basically the "boring but works" stack, while also being more performant and with less configuration required.
I guess I'm just an old man pumping my fist and yelling at the clouds at this point, but I think compartmentalizing every possible aspect of development and then making each one of those as complex as possible lead AI in the bureaucratic deterioration of the trade.
I love Vite, Vitest, Oxlint and Oxfmt and look in their direction for most of my new projects! I hope these folks manage to get a bunch of money and can fund the continued development for at least the next decade.
Sure beats opening some ancient project and seeing some mix of Gulp, Grunt, webpack and a bunch of other disjointed stuff (I migrated that one over to also use the newer stack).
> I hope these folks manage to get a bunch of money and can fund the continued development for at least the next decade.
I believe VoidZero has been acquired by Cloudflare [1], so money should not be an issue. Question is if Cloudflare will be willing to continue letting these people work on Vite and Vite+ features that benefit all cloud platforms, not just Cloudflare.
Making all this (for example) work nicely together can be tricky: Vite, ESLint, Prettier, Typescript and React, especially if it's full stack with SSR.
If you only focus on the front-end and remove Typescript from the equation it becomes easy enough. We'll have to see if Vite+ helps for the more complex cases.
> Sure beats opening some ancient project and seeing some mix of Vite, Vitest, Oxlint and Oxfmt and a bunch of other disjointed stuff (I migrated that one over to also use the newer stack).
I mean if I see those in N years, I'll be happier than with the older stack that came before them - the jank levels seem to generally be decreasing with every next attempt to get things right!
Vite had five major version in the four years 2022-2026. Version 3 => 4 => 5 => 6 => 7 => 8. Each one of those had breaking changes and required devs to go through a migration. It's too much. And for what? It's not as if it is dramatically better now than it was in version 3.
I can't say I would really look forward to bringing this level of needless churn and constant disruption to the rest of my development toolchain. Anyway, Vite+ is really just wrapping existing tools into an abstracted command-line interface? And so I have more layers of indirection to wade through in order to get the thing to do what I want? So far I am not optimistic about this prospect...
I've followed all the main migrations and I've say they where really quite smooth, can't remember having any major issues and each time it tended to be worth it.
Because VILT is dead and Livewire is now on version 1100?
I've worked on both stacks in the last few years across several clients. Honestly like with anything in tech it seems to mostly fall apart with half-regarded usage of the tools in growing teams that don't care about their quality in favour of "get ticket done".
In my experience the bundling isn’t really the important aspect (though it also doesn’t harm anything), it’s more just having an ecosystem of plugins for code transpiling, static asset inclusion (e.g. text files) etc and a configuration format folks are already used to.
For me, the main benefit is deployment bundle/artifact size reduction. Mostly from dropping unneeded files from node_modules. Many packages include both esm and cjs builds, sources, docs, TS types, etc. stuff that you don’t need in prod.
This matters for lambdas, for example, because deployed code size has limits there.
Running typescript without compilation is still tricky with plain Node. `vite dev` has amazing DX not available for Node programs. I'm wondering if Vite+ tackles this problem.
Don't we have `tsx` and `nodemon` (or the native Node reloader) for that? What are the DX gaps you see on the server side out of on-the-fly transpilation and reload on watch?
Yes, I use tsx for Node programs. It's not great when sharing the same codebase for both client and server code, they have completely different dev workflows.
In theory, typescript doesn't need to be transpiled, you can run ts files using `node --experimental-strip-types file.ts` as long as you don't use any code that needs transpilation (like typescript enums).
No, because of ESM import resolution rules. Typescript suggests extensionless imports, making it incompartible with ESM and therefore Node. Luckly, `node --import=tsx file.ts` handles imports well.
This is especially hairy when making a typescript library that is distributed non-compiled (without dist/) and is supposed to run in both browser and Node.
This allows you to use fully-specified imports in TypeScript files, which works basically everywhere — NodeJS, TypeScript, bundlers, etc.
The exceptions are browsers (obviously, only normal JS syntax there), and packages inside `node_modules`, which NodeJS will not do any type stripping for. So if you're writing a library, you'll probably still need to distribute the compiled sources, rather than distributing the raw TypeScript files alone. Or you use the JSDoc syntax for TypeScript, which can do everything that .ts files can do, but is more verbose and idiosyncratic.
One advantage of precompilation is risk reduction. Say tsx gets hacked somehow (hardly unprecedented with Node modules!) you’ve got it running on your production server exposed to the internet. Precompilation on a CI pipeline is still a risk but a significantly lower one.
I’ve found if you want interop ts esm and js cjs you need to compile your code - and then `tsc` doesn’t bundle your dependencies for you and outputs incomplete code.
In cases where startup time matters and you have a slow disk, bundling can drastically reduce the number of filesystem calls you have to make for large dependency graphs.
Yeah, not only the name: they’re also going with various semiotic signs that are strongly associated with a subscription service, including their website design, choice of typography, and even the press release–style copy.
Looks like they have been acquired by Cloudflare, and pivoted to fully open source, but they haven’t really tweaked their messaging to make that fully land with unsuspecting visitors.
It’s kinda like the reverse situation of open source projects that switch to a source available license, but keep the aesthetics of an open source project. Kinda funny!
This is just what modern languages have out of box. (Like rust and go.) it’s a true shame that web isn’t actually unified behind a type safe language with a single solid toolchain. It’s a huge pain to manage and I’m curious how much money it’s cost the industry. “Vite+” isn’t a true solution to that. There are many competing toolchains. And no default standardized one.
I'm not very familiar with Rust, but doesn't cargo pull a lot of external dependencies for most projects? I really like how Go can do everything with just the standard library, but I wasn't aware Rust was similar. For typescript we've moved our stuff to bun. It has it's own risk management perspective compared to node, but at least it's now possible to build web services without having to rely on a bunch of external dependencies. Which in our highly regulated business would require security policies for each dependency explaining the risks, why we accept them and how we mitigate them.
> without having to rely on a bunch of external dependencies. Which in our highly regulated business would require security policies for each dependency explaining the risks, why we accept them and how we mitigate them.
How about the dependencies Bun is pulling? How did you ever managed to pass security policies with Bun which has so many segfaults that nobody even bothers to write CVEs for them.
Cargo itself doesn't pull the dependencies, but yes to Rust's standard library being much more lean than Go. Bring your own HTTP, text templating, and such, but core data structures are provided.
Go gives you a bunch of goodies in the standard library.
Rust provides things like your build system, testing, and package management all together, which is what I assume OP meant.
Vite+ isn't a layer, it's "just" a high-performance suite of excellent tools that work well together to provide a great DX for developers.
Vite+ can improve and simplify what developers are already doing with ad-hoc collections of tools. Vite is already an industry standard, and Vite+ has a good chance of achieving that status as well.
Some barbarian without a grasp of the Dutch language knee-jerked the down-vote button so I'll add a Swedish version which adds an important attribute.
Imse vimse spindel klättra upp för trå'n.
Ner faller regnet, spolar spindeln bort.
Upp stiger solen, torkar bort allt regn.
Imse vimse spindel klättrar upp igen.
Here's how to interpret this saga of the ever-climbing little spider in the context of web development. It climbs up its tread (klättra upp för trå'n) 'cause that new framework will sure make catching those flies (clicks/jobs/likes/whatevers) easier. And then the rain starts (the CVEs start piling up, the corrupted packages come flooding in) and the hapless spider gets thrown off its web (Pwned!) until the sun comes back and dries away the rain (a new framework, yay, this will solve all problems) upon which the spider climbs up its thread again.
It's all great to leverage until something breaks in a middle layer and you can't reproduce without submitting your entire project in the GitHub issue.
I think web development does not need that many layers. Usually there is a clear purpose for each layer. I think most problems in web are self-created.
Is everyone project so simple that it can fit in these "vp check" / "vp dev" commands? Like even for my amateurish web app, I have a custom web server with a self-signed certificate with an "/etc/hosts" domain; and for checks I need to do custom checks for GraphQL and a couple of cloned NPM packages.
I have been doing "modern web" things since essentially day zero (you kids with your fancy JIT compiled javascript interpreters!)
SvelteKit, and by extension, Vite, has been the single most productive webstack I have ever used. If this offers anything on top of that, I welcome it with open arms.
I have removed vite because dev build and reload is noticable slower than just esbuild and browser refresh. Vite does nothing for me that an LLM can not just trivially rebuild in a bespoke manner.
I am actually pushing our frontend devs to remove more and more dependencies and leverage LLMs to just write the code instead of all the dumbass packages in hellscape of supply chain attacks via node/npm.
You are signing up for another hellscape of unmaintainable slop. Enable package cooldowns and only whitelist internal packages and you are better off than 90%
How do you bundle web workers that import dependencies? iirc the issue in esbuild for that is still open and users are manually building their workers as separate entry points, which is very fragile.
I am a big fan of Vite. But I have zero clue what those other tools are. I swear to God, I just put my head down to do some work and all the sudden, frontend tooling has evolved. I wonder if there is a push towards a "boring but works" stack.
the other tools are for testing, bundling, linting and formatting. Previously you would use different tools from very different open source projects for these things, with different configurations, update cycles etc. Now it's all covered by one simple toolchain. Vite+ is basically the "boring but works" stack, while also being more performant and with less configuration required.
That's what I liked from Bun's proposition. A single binary that just works. Hopefully the others take notice.
This is the latest emerging "boring but works" stack.
I guess I'm just an old man pumping my fist and yelling at the clouds at this point, but I think compartmentalizing every possible aspect of development and then making each one of those as complex as possible lead AI in the bureaucratic deterioration of the trade.
I love Vite, Vitest, Oxlint and Oxfmt and look in their direction for most of my new projects! I hope these folks manage to get a bunch of money and can fund the continued development for at least the next decade.
Sure beats opening some ancient project and seeing some mix of Gulp, Grunt, webpack and a bunch of other disjointed stuff (I migrated that one over to also use the newer stack).
> I hope these folks manage to get a bunch of money and can fund the continued development for at least the next decade.
I believe VoidZero has been acquired by Cloudflare [1], so money should not be an issue. Question is if Cloudflare will be willing to continue letting these people work on Vite and Vite+ features that benefit all cloud platforms, not just Cloudflare.
1. https://blog.cloudflare.com/voidzero-joins-cloudflare/
Making all this (for example) work nicely together can be tricky: Vite, ESLint, Prettier, Typescript and React, especially if it's full stack with SSR.
If you only focus on the front-end and remove Typescript from the equation it becomes easy enough. We'll have to see if Vite+ helps for the more complex cases.
> Sure beats opening some ancient project and seeing some mix of Vite, Vitest, Oxlint and Oxfmt and a bunch of other disjointed stuff (I migrated that one over to also use the newer stack).
I mean if I see those in N years, I'll be happier than with the older stack that came before them - the jank levels seem to generally be decreasing with every next attempt to get things right!
Vite had five major version in the four years 2022-2026. Version 3 => 4 => 5 => 6 => 7 => 8. Each one of those had breaking changes and required devs to go through a migration. It's too much. And for what? It's not as if it is dramatically better now than it was in version 3.
I can't say I would really look forward to bringing this level of needless churn and constant disruption to the rest of my development toolchain. Anyway, Vite+ is really just wrapping existing tools into an abstracted command-line interface? And so I have more layers of indirection to wade through in order to get the thing to do what I want? So far I am not optimistic about this prospect...
Echoing another person, but we did all those major version migrations without any breakage / churn. What were the breaks you ran into?
I've followed all the main migrations and I've say they where really quite smooth, can't remember having any major issues and each time it tended to be worth it.
The churn is the product.
Truly have so much trouble keeping up with the frontend (or JavaScript?) ecosystem. I so miss working in laravel. Wish more jobs paid well to use it.
I feel ya, we're slowly phasing out our Laravel monolith for python lambdas. I miss those beautiful Laravel 6 days!
Trust me you don't want to work with Laravel Livewire and Alpine.js, that would still require keeping up and for a less than satisfying result.
Because VILT is dead and Livewire is now on version 1100?
I've worked on both stacks in the last few years across several clients. Honestly like with anything in tech it seems to mostly fall apart with half-regarded usage of the tools in growing teams that don't care about their quality in favour of "get ticket done".
you actually don't have to keep up, whatever you were using still works
Toolchain Grand Vitesse
it worked for uv so i can imagine a competent team can do the same thing for javascript!
Surprised to see this is the only uv reference in the comments!
Feels like an obvious comparison to me, and a very welcome development for the JS ecosystem.
uv made me actually _enjoy_ working in Python again.
Can it be used for Node builds or browser-only same as Vite?
It uses Vite so the same limitations as Vite. However, I have been using Vite for my NestJS servers without any problem with `vite-plugin-node`. See example at https://github.com/leosuncin/nest-vite-example/blob/master/v...
Here is my particular incantation for targeting node that is working very well: https://pastebin.com/ynz4B5X0
Essentially you pretend to be a library
I am using Vite+ for CLIs as well, yes. You don't use Vite as dev server then but lint, format, task running and caching is still there!
I'd be interested in seeing this implementation if it's publicly available. Do you have a GitHub link? Thanks!
I'm always curious of the use case when someone proposes Node code bundling. What's the advantage? Obfuscation in SEA?
In my experience the bundling isn’t really the important aspect (though it also doesn’t harm anything), it’s more just having an ecosystem of plugins for code transpiling, static asset inclusion (e.g. text files) etc and a configuration format folks are already used to.
For me, the main benefit is deployment bundle/artifact size reduction. Mostly from dropping unneeded files from node_modules. Many packages include both esm and cjs builds, sources, docs, TS types, etc. stuff that you don’t need in prod. This matters for lambdas, for example, because deployed code size has limits there.
Running typescript without compilation is still tricky with plain Node. `vite dev` has amazing DX not available for Node programs. I'm wondering if Vite+ tackles this problem.
Don't we have `tsx` and `nodemon` (or the native Node reloader) for that? What are the DX gaps you see on the server side out of on-the-fly transpilation and reload on watch?
Yes, I use tsx for Node programs. It's not great when sharing the same codebase for both client and server code, they have completely different dev workflows.
In theory, typescript doesn't need to be transpiled, you can run ts files using `node --experimental-strip-types file.ts` as long as you don't use any code that needs transpilation (like typescript enums).
Still need tsx to do type checking
No, because of ESM import resolution rules. Typescript suggests extensionless imports, making it incompartible with ESM and therefore Node. Luckly, `node --import=tsx file.ts` handles imports well.
This is especially hairy when making a typescript library that is distributed non-compiled (without dist/) and is supposed to run in both browser and Node.
https://github.com/nodejs/node/issues/46006
https://github.com/microsoft/TypeScript/issues/16577
You can use one of the following:
`allowImportingTsExtensions: true` (https://www.typescriptlang.org/tsconfig/#allowImportingTsExt..., useful if you're running `tsc` in noEmit mode as a linter)
`rewriteRelativeImportExtensions: true` (https://www.typescriptlang.org/tsconfig/#rewriteRelativeImpo..., useful if you're using `tsc` to compile TS files to JS.
This allows you to use fully-specified imports in TypeScript files, which works basically everywhere — NodeJS, TypeScript, bundlers, etc.
The exceptions are browsers (obviously, only normal JS syntax there), and packages inside `node_modules`, which NodeJS will not do any type stripping for. So if you're writing a library, you'll probably still need to distribute the compiled sources, rather than distributing the raw TypeScript files alone. Or you use the JSDoc syntax for TypeScript, which can do everything that .ts files can do, but is more verbose and idiosyncratic.
One advantage of precompilation is risk reduction. Say tsx gets hacked somehow (hardly unprecedented with Node modules!) you’ve got it running on your production server exposed to the internet. Precompilation on a CI pipeline is still a risk but a significantly lower one.
If only the whole JavaScript wasn't as dependency hell of single function packages....
@afavour if you need precompilation in CI can't you simply use... tsc?
I’ve found if you want interop ts esm and js cjs you need to compile your code - and then `tsc` doesn’t bundle your dependencies for you and outputs incomplete code.
In cases where startup time matters and you have a slow disk, bundling can drastically reduce the number of filesystem calls you have to make for large dependency graphs.
Is there a subscription with this?
I'm just wary about anything with a '+' and I assume there is a subscription attached to it.
Looking at this it doesn't look like it.
My first thought too. "$name+" is strongly coded now as "subscription service for $name"
Yeah, not only the name: they’re also going with various semiotic signs that are strongly associated with a subscription service, including their website design, choice of typography, and even the press release–style copy.
Looks like they have been acquired by Cloudflare, and pivoted to fully open source, but they haven’t really tweaked their messaging to make that fully land with unsuspecting visitors.
It’s kinda like the reverse situation of open source projects that switch to a source available license, but keep the aesthetics of an open source project. Kinda funny!
Says:
"It is fully open source under the MIT license"
I think that used to be the idea but then they got acqui-hired
Naming is worrisome!
Layer on layer on layer on layer on layer.... Web development is just a meme by now
This is just what modern languages have out of box. (Like rust and go.) it’s a true shame that web isn’t actually unified behind a type safe language with a single solid toolchain. It’s a huge pain to manage and I’m curious how much money it’s cost the industry. “Vite+” isn’t a true solution to that. There are many competing toolchains. And no default standardized one.
I'm not very familiar with Rust, but doesn't cargo pull a lot of external dependencies for most projects? I really like how Go can do everything with just the standard library, but I wasn't aware Rust was similar. For typescript we've moved our stuff to bun. It has it's own risk management perspective compared to node, but at least it's now possible to build web services without having to rely on a bunch of external dependencies. Which in our highly regulated business would require security policies for each dependency explaining the risks, why we accept them and how we mitigate them.
> without having to rely on a bunch of external dependencies. Which in our highly regulated business would require security policies for each dependency explaining the risks, why we accept them and how we mitigate them.
How about the dependencies Bun is pulling? How did you ever managed to pass security policies with Bun which has so many segfaults that nobody even bothers to write CVEs for them.
Cargo itself doesn't pull the dependencies, but yes to Rust's standard library being much more lean than Go. Bring your own HTTP, text templating, and such, but core data structures are provided.
Go gives you a bunch of goodies in the standard library.
Rust provides things like your build system, testing, and package management all together, which is what I assume OP meant.
Vite+ isn't a layer, it's "just" a high-performance suite of excellent tools that work well together to provide a great DX for developers.
Vite+ can improve and simplify what developers are already doing with ad-hoc collections of tools. Vite is already an industry standard, and Vite+ has a good chance of achieving that status as well.
Deze vuist op deze vuist. Deze vuist op deze vuist. Deze vuist op deze vuist. En zo klim ik naar boven.
You probably need to see a video or gif to get it.
Some barbarian without a grasp of the Dutch language knee-jerked the down-vote button so I'll add a Swedish version which adds an important attribute.
Here's how to interpret this saga of the ever-climbing little spider in the context of web development. It climbs up its tread (klättra upp för trå'n) 'cause that new framework will sure make catching those flies (clicks/jobs/likes/whatevers) easier. And then the rain starts (the CVEs start piling up, the corrupted packages come flooding in) and the hapless spider gets thrown off its web (Pwned!) until the sun comes back and dries away the rain (a new framework, yay, this will solve all problems) upon which the spider climbs up its thread again.It's all great to leverage until something breaks in a middle layer and you can't reproduce without submitting your entire project in the GitHub issue.
Pretty much all software is built like that.
I think web development does not need that many layers. Usually there is a clear purpose for each layer. I think most problems in web are self-created.
Is everyone project so simple that it can fit in these "vp check" / "vp dev" commands? Like even for my amateurish web app, I have a custom web server with a self-signed certificate with an "/etc/hosts" domain; and for checks I need to do custom checks for GraphQL and a couple of cloned NPM packages.
Don't be so negative nancy here!
I have been doing "modern web" things since essentially day zero (you kids with your fancy JIT compiled javascript interpreters!)
SvelteKit, and by extension, Vite, has been the single most productive webstack I have ever used. If this offers anything on top of that, I welcome it with open arms.
Far from being a meme!
vp + sv seems to work very well, when I tried it. And oxfmt supports svelte now too!
Doug McIlroy once said: "Make each program do one thing well".
Then he'd love this. Like Unix, Vite+ is a collection of programs that do one thing well.
Im tried boss
It’s a great move for Cloudflare to have bought up voidzero.
I appreciate the effort to bring things together in this but…
> Vite+ will manage your global Node.js runtime and package manager.
What? Why?
You’re really going all-in if you adopt this; and… for what? A bit of cozy tooling around existing standard ways of doing things?
Ok, sure; I like tools, like vite.
…but even for an opinionated tool, this is extraordinarily opinionated. Like next.js
Im skeptical.
The pitch of bringing things together seems strong, but did we go too far here?
Reading reviews of people using this didn't really convince me.
It seems to be running on the coat tails of the vite name, rather than its own merit.
I have removed vite because dev build and reload is noticable slower than just esbuild and browser refresh. Vite does nothing for me that an LLM can not just trivially rebuild in a bespoke manner.
YMMV
I am actually pushing our frontend devs to remove more and more dependencies and leverage LLMs to just write the code instead of all the dumbass packages in hellscape of supply chain attacks via node/npm.
You are signing up for another hellscape of unmaintainable slop. Enable package cooldowns and only whitelist internal packages and you are better off than 90%
How do you bundle web workers that import dependencies? iirc the issue in esbuild for that is still open and users are manually building their workers as separate entry points, which is very fragile.