I dont think thats much of a debate. And I say that as a Vue fanatic. Altough, you could argue that since Vite was supposed to be for vue, therfore any impact vite has is by extension Vue`s impact. And I guess from that point of view vue has had more of an impact because Vue's impact encompases all of vite's impact pluss its own impact in other areas.
Nikhil has spoken about how he's also looking at possibly using Vite's Environment API to reduce the Vinxi magic and rather turn it more into an orchestration tool. Edit: We (at TanStack Start) aren't looking to drop Vinxi anytime soon. Evan has hinted at there being a closer integration between Vite and Nitro that may be coming in the future, which is probably the only avenue of exploration we'd make if required.
Bun + Vite + Svelte are the only reason why I go near Front-End Development. Especially Vite. Now I can spend more time actually learning TS than fighting with some random issue.
honestly I don't know whether it's Bun or Vite doing the magic but not having to know that is nice too, it's been such a pain to make tsc and node happy!
I purposely built two different projects with Next and Remix respectively just to see the difference. I was blown away with how easy and fast Remix is. Next works, it has RSC, and Turbopack is a nice touch, but I'm honestly fine with the simplicity and speed that Remix provides. Also, using Context to support useLoaderData and useGetLoaderData is wonderful.
I laughed like a maniac when you said "almost regretting this Rust rewrite", as even after two years of Rust programming, I still think the same every time I do "cargo build". (watching this as I wait for my workspaced-project to fully compile)
just soämäe clarification, H3 is part of the unjs echosystem and is a minimal backend li rary e.g. Express, nitro is built upon it and is a backend framework, it adds deployment configs, db connectors, caching etc. Vinxi is built on top of Nitro to have multiple things. But H3 is an integral part of Nitro (nitro doesn't work without it) and Nitro is an integral part of vinxi.
Hey Theo, you said you think Storybook sucks... why? For the technical setup? For the functionality? Or even the whole idea? Any alternatives you'd recommend?
@@snowe.. Thanks for your reply. Yeah, I know Theo does that a lot. It's not that I'm changing my professional decision made for the company I'm working for years ago because of a side comment in a TH-cam video. Not a fan of hype engineering. Business uses the tool to give feedback early, devs use it a lot for a11y and more. Works great for us. But I love to hear different opinions and views about things. We can always learn. And that's why I always ask why in such cases.
@ gotcha gotcha. Yeah good job there, you said it in a very neutral tone so it seemed (to me) like you were basing decisions on what Theo says. Clearly I was wrong. 👍🏼
When I was looking into analog I got really deep into learning about nitro. The reason I didn't use analog was that because it didn't properly support nitro hooks. Love the unjs stuff
SSR is such a pain in the current Vite I just skipped it on our projects (or used NEXT) but would love to use SSR if it becomes easily useable on Vite.
I * really * like Vite...but migrating a large project to it was really difficult and was scrapped in the end. After a while, we tried again, with rsbuild - did it in 2 days, there's barely any config, and the HMR and build speed still astonish us daily, months later...
@@ajithav2011 For us, it was: a) incompatibility with some of the libraries we're using - primarily Antd 3.x and b) how "hard-fixed" some of the defaults are. (Antd is a separate story, as we migrated to 5.x, only to find out first-hand about horrible performance issues that there is no resolution for until 6.x, so had to stay with 3), With Vite, and going for least configuration, we needed to do a lot of work in the project to overcome the errors and warnings - admittedly, it's good defaults, and most of our problems came from doing non-standard stuff that use of webpack and plugins introduced over time, but it wasn't easy to do gradually - more of a all or nothing approach. Compared to that, with rsbuild, we got even more performance (frontend build and deploy went from 1.9 min with webpack to 10 sec total with rsbuild) , equal ability to just remove 20+ dev packages of different webpack and babel plugins, and the migration was just less painful / more gradual...We got it working very quickly, and then polished. For Vite, it was 90% of work needed up-front, just to get the app running. I think a lot of it is project/situation-specific, and if starting with a fresh project, it's a great choice - I really like the overall feel of it. But rsbuild gives no reason to be avoided, and comes with some strong points on top as well.
Same. I tested both Vite and RSBuild. Ended up switching a large CRA app to RSBuild w ease. Some things I noticed: Vite had problems w file extensions. Like having JSX in a .js file. RSBuild is crazy fast in starting the dev server, hmr, and building a dist. Both are great imo. But one worked better for our use case. And we still use ViteTest for testing.
@@ajithav2011 I'll try again shorter. A) Incompatibility with some of the libraries we're using - like Antd 3.x B) A lot of work to overcome errors and warnings - defaults are really good, but if going for least configuration, kind of hard-wired, so it's difficult to work around them.
Since I do not use vercel I probably lack understanding on how easy it is to deploy it. My solution is just to have git repo on my prod server for my small website projects i Just run the script that pulls in latest, builds out and reboots the server, not idea but what is there something vercel specific that scripting thing yourself provide?
SSR still seems mostly useless to me and massively over-hyped. The complexity it adds is extreme as well. Can someone explain why on earth SPA architecture isn't still preferable?
@@nickmurdaugh9856As someone who works on internal company web applications, if that's the only benefit, I'm pretty annoyed everything is moving towards the SSR route.
What's so extremely complicated about it? It surely add complexity but like every new paradigm you just need to get used to it. 99% of the times you just have to mindful of not calling browser APIs outside of useEffect or check if you are in the browser or not before calling it
Depends on what you are building. Sometimes you need to generate the base html rather than serve it statically, conditionally adding head tags based on region, AB testing your assets. Etc somethings are just cleaner and easier when you can dynamically generate the first thing a user sees Basic SSR for initial page render is not hard or complex. It’s server components etc that make it difficult
Hearing Theo talk about SSR just confuses me, since rendering on the server side has existed for soooo long. Can the frontend SPA guys treat it more like yeah we value SSR too and want it in our new tech-stack.
Awesome project, but It's hard to get attention these days if you don't have enough budget for marketing or if that's about a chinese project, am I exaggerating?
I'm a project manager do some coding but not really good. Sometimes on my free time I watch this channel to try to learn. This video was way to advance for me 😅 could have used diagrams and examples but thanks for the work
I don't get why you'd say vite doesn't work on server... Sveltekit is bundled with Vite. I've personally had great success with Vite to build my own Svelte server islands architecture. Where is the substance behind this claim?
I really like Vite, never had much of an issue with it but "it's a standard for a reason" is a bad argument because it's basically the same argument for Webpack.
just open the first page of the Vite docs, it's literally the first sentence, they even have a recording of a proper pronunciation and it's nothing new, they have it there since day one
Not sure why people need all these tools. I've been building with HTML, CSS and vanilla JS in Notepad++ for the last 2 years, and never found a need for any frameworks etc. Admittedly, stuff takes longer but that's ok for me. I only land up with 3 files max to edit.
There's too many frameworks for basically doing the same thing, it's nice there are different options to choose from, but it just feels as if though people just keep creating new frameworks for the sake of making new frameworks around certain things, when one or two already exists and does most of or all of what they want already by copying the same features of the ones that already exist, but with little extra things added in that they want to see. Having too many different frameworks just complicates things, and if you ever try to step away from one to move to another, you end up having to change a lot of your stuff to convert to the new stuff.
Would it be fair to say that Vite has had almost more of an impact in the web-dev world than Vue has?
definitely it would be. idek why vue has so many stars on GitHub I honestly think vite should have way more
Significantly more impact than Vue
yes
I dont think thats much of a debate. And I say that as a Vue fanatic. Altough, you could argue that since Vite was supposed to be for vue, therfore any impact vite has is by extension Vue`s impact. And I guess from that point of view vue has had more of an impact because Vue's impact encompases all of vite's impact pluss its own impact in other areas.
By far yeah
"SSR is the future!"
...so what did I unlearn my PHP for?! Damnit! /s
could've just bundled a php interpreter and wrote php clientside
@@IngwiePhoenix_nb Two hundred requests per second!!!
I CALL IT VITE, NOT VIT, I DO NOT CARE.
@@SIGMA_BLYATlearn to pronounce properly
PHP never went away, with good reason.
UnJS, Nitro, Vite, H3, Vitest ...all came from the Vue developers. Just sayin'. Vue and Nuxt are fantastic. I'd like to see more people using them.
I recently started using Nuxt and love it so far
Nikhil has spoken about how he's also looking at possibly using Vite's Environment API to reduce the Vinxi magic and rather turn it more into an orchestration tool.
Edit: We (at TanStack Start) aren't looking to drop Vinxi anytime soon. Evan has hinted at there being a closer integration between Vite and Nitro that may be coming in the future, which is probably the only avenue of exploration we'd make if required.
Bun + Vite + Svelte are the only reason why I go near Front-End Development.
Especially Vite. Now I can spend more time actually learning TS than fighting with some random issue.
honestly I don't know whether it's Bun or Vite doing the magic but not having to know that is nice too, it's been such a pain to make tsc and node happy!
Next.js not using vite is starting to really frustrate me. It would make maintaining a monorepo easier.
easy fix, stop using nextjs and start using tanstack start
use REMIX
Use Nuxt (Vite + Nitro + h3 under the hood - not sure why this wasn't mentioned in the vid despite all the apparent love for this stack).
Turbopack became almost like a meme, rspack and rolldown move faster than this
I purposely built two different projects with Next and Remix respectively just to see the difference.
I was blown away with how easy and fast Remix is. Next works, it has RSC, and Turbopack is a nice touch, but I'm honestly fine with the simplicity and speed that Remix provides. Also, using Context to support useLoaderData and useGetLoaderData is wonderful.
this blacksmith sponsorship is one of the VERY few ads that I actually checked out, and do actually like! thank you!
When will we get an in-depth UnJS/Nitro video 👀
I laughed like a maniac when you said "almost regretting this Rust rewrite", as even after two years of Rust programming, I still think the same every time I do "cargo build".
(watching this as I wait for my workspaced-project to fully compile)
PHP: look what they have to do to mimic a fraction of my power
I was listening passively, and suddenly I heard Theo say, "Vladimir has been killing it too."
lol. apologies for my sense of humor
I wrote a few projects at my job with Snowpack and quickly moved them all over to Vite
just soämäe clarification, H3 is part of the unjs echosystem and is a minimal backend li rary e.g. Express, nitro is built upon it and is a backend framework, it adds deployment configs, db connectors, caching etc. Vinxi is built on top of Nitro to have multiple things. But H3 is an integral part of Nitro (nitro doesn't work without it) and Nitro is an integral part of vinxi.
1:01 Ben debugging build issues over the phone while on the toilet. Legend
Frankly speaking, I still do tag and manually src my JS files, like we did 20 years ago, and it just works.
Am I the only one waiting for characters in Theo's skit to develop their lore and backstory into a full cinematic universe?
Came for article stayed for the ad 🖤
I swear i was looking onto cool ways to implement server side stuff into Vite literally 30 seconds before opening the video
I liked that blacksmith ad lol
Huh. Only recently heard of Vite, and now I’m excited.
Hey Theo, you said you think Storybook sucks... why?
For the technical setup? For the functionality? Or even the whole idea?
Any alternatives you'd recommend?
Theo says a bunch of stuff sucks when it doesn't, and says a bunch of stuff is good when it isn't. If you like storybook then use storybook.
@@snowe.. Thanks for your reply. Yeah, I know Theo does that a lot.
It's not that I'm changing my professional decision made for the company I'm working for years ago because of a side comment in a TH-cam video.
Not a fan of hype engineering.
Business uses the tool to give feedback early, devs use it a lot for a11y and more.
Works great for us.
But I love to hear different opinions and views about things.
We can always learn. And that's why I always ask why in such cases.
@ gotcha gotcha. Yeah good job there, you said it in a very neutral tone so it seemed (to me) like you were basing decisions on what Theo says. Clearly I was wrong. 👍🏼
0:18 Snowpack was first called "Pika", Snowpack & Vite initially both used esbuild internally.
I think it's time that you should try out nuxt ;)
Vite is the best bundler!
- Software engineer with 10+ years experience
When I was looking into analog I got really deep into learning about nitro. The reason I didn't use analog was that because it didn't properly support nitro hooks. Love the unjs stuff
I’m excited for nuxt 4, it will leverage all these changes
SSR is such a pain in the current Vite I just skipped it on our projects (or used NEXT) but would love to use SSR if it becomes easily useable on Vite.
Half of things is going above my mind but i am still watching
3:48 bro just casually mentions this cool tool, we need a monthly video for just cool stuff that we might have missed or never heard of
I * really * like Vite...but migrating a large project to it was really difficult and was scrapped in the end. After a while, we tried again, with rsbuild - did it in 2 days, there's barely any config, and the HMR and build speed still astonish us daily, months later...
I am in a similar situation. What were the issues you faced when migrating to vite?
@@ajithav2011 For us, it was: a) incompatibility with some of the libraries we're using - primarily Antd 3.x and b) how "hard-fixed" some of the defaults are.
(Antd is a separate story, as we migrated to 5.x, only to find out first-hand about horrible performance issues that there is no resolution for until 6.x, so had to stay with 3),
With Vite, and going for least configuration, we needed to do a lot of work in the project to overcome the errors and warnings - admittedly, it's good defaults, and most of our problems came from doing non-standard stuff that use of webpack and plugins introduced over time, but it wasn't easy to do gradually - more of a all or nothing approach.
Compared to that, with rsbuild, we got even more performance (frontend build and deploy went from 1.9 min with webpack to 10 sec total with rsbuild) , equal ability to just remove 20+ dev packages of different webpack and babel plugins, and the migration was just less painful / more gradual...We got it working very quickly, and then polished. For Vite, it was 90% of work needed up-front, just to get the app running.
I think a lot of it is project/situation-specific, and if starting with a fresh project, it's a great choice - I really like the overall feel of it. But rsbuild gives no reason to be avoided, and comes with some strong points on top as well.
Same. I tested both Vite and RSBuild. Ended up switching a large CRA app to RSBuild w ease.
Some things I noticed:
Vite had problems w file extensions. Like having JSX in a .js file.
RSBuild is crazy fast in starting the dev server, hmr, and building a dist.
Both are great imo. But one worked better for our use case. And we still use ViteTest for testing.
@@ajithav2011 Not sure why my detailed reply was removed...
@@ajithav2011 I'll try again shorter. A) Incompatibility with some of the libraries we're using - like Antd 3.x B) A lot of work to overcome errors and warnings - defaults are really good, but if going for least configuration, kind of hard-wired, so it's difficult to work around them.
Very nice sketch for the ad lol
SSR will be "more and more the future" if and when you are able to render more than 200 pages per second.
Your advertisements remind me of Saul Goodman's commercials.
Playing around with Nitro and Vinxi, something is comming. It's not what you think.
in my company we use rsbuild and it is great
Since I do not use vercel I probably lack understanding on how easy it is to deploy it. My solution is just to have git repo on my prod server for my small website projects i Just run the script that pulls in latest, builds out and reboots the server, not idea but what is there something vercel specific that scripting thing yourself provide?
I've never been this early before
Same
SSR still seems mostly useless to me and massively over-hyped. The complexity it adds is extreme as well. Can someone explain why on earth SPA architecture isn't still preferable?
Seo.
It can also lead to faster load times for a lot of content.
@@nickmurdaugh9856As someone who works on internal company web applications, if that's the only benefit, I'm pretty annoyed everything is moving towards the SSR route.
Without JavaScript your website is not running.
Faster load times, and SEO friendly
For an easy SSR, I use Nuxt and it's amazing how good it is!
What's so extremely complicated about it? It surely add complexity but like every new paradigm you just need to get used to it.
99% of the times you just have to mindful of not calling browser APIs outside of useEffect or check if you are in the browser or not before calling it
Depends on what you are building. Sometimes you need to generate the base html rather than serve it statically, conditionally adding head tags based on region, AB testing your assets. Etc somethings are just cleaner and easier when you can dynamically generate the first thing a user sees
Basic SSR for initial page render is not hard or complex. It’s server components etc that make it difficult
How your github is showing file icons ? how, what did you do? I like that
Hearing Theo talk about SSR just confuses me, since rendering on the server side has existed for soooo long. Can the frontend SPA guys treat it more like yeah we value SSR too and want it in our new tech-stack.
Thank you for the video
I hope you talk about farmfe
Awesome project, but It's hard to get attention these days if you don't have enough budget for marketing or if that's about a chinese project, am I exaggerating?
Theo, can you influence vercel to move next 16 to vite?
it will be "the standard" for approximately n years before something else comes along
Hope Rolldown will support SystemJS or atleast allow to continue to use Rollup with Vite under the hood if they don't.
I did try snow pack, the dev/release differences pointed the way forward!!
Ok the addspot was funny for Rust and true.
I'm a project manager do some coding but not really good. Sometimes on my free time I watch this channel to try to learn. This video was way to advance for me 😅 could have used diagrams and examples but thanks for the work
out of curiosity what is wrong with storybook?
The explanation about SSR got me thinking, isn't this just PHP?
Still just running express docker and static front :)
No "body tag" in twitch 😊?
"Storybook kinda sucks" is a... strong take.
Why does it?
It had 669 likes .. it's so pleasing to push it to 670 😂🎉
I thought isomorphic rendering is solved by now, but it seems it isn't.
Btw Mithril community solved this already 10 years ago
If Storybook sucks, what would you use instead?
And no promised out of the box federation solution 😭
six point, ohhhhhh let's friggin go~!
video about waku!
BTW React router v7 has been released and it's replacing Remix. I'm surprised you haven't covered this.
I covered it months ago?
There’s vike though, plugins into vite
"Other than getting off that old way of building sooner rather than later".
No. I dont think I will.
Thx, One JS / Tamagui
I don't get why you'd say vite doesn't work on server... Sveltekit is bundled with Vite. I've personally had great success with Vite to build my own Svelte server islands architecture. Where is the substance behind this claim?
snowpack gang!
Just asking, you read from a script right?
I'm using it right now
Let's refactor gulp in rust (=
I really like Vite, never had much of an issue with it but "it's a standard for a reason" is a bad argument because it's basically the same argument for Webpack.
It was the same argument for webpack back in the day. And during those times it was as true for webpack as it now is for vite.
Most things humans do are for a reason. Doesn't have to be a good reason.
Vinxi is pronounced like vinchi (to rhyme with da Vinci)
SSR is basically a modern way to say PHP + JQuery
No it's not. It really isn't.
All of this just to be php?
Make video on rspack
T3 is gone now u have a better type safety stack. Vit6, tanstack start and tailwind 4
Fresh off the printers
Vite 6??? I'm still at vite 1
thats more of a brag than a fun fact :P
Theo?
Btw, the Storybook has changed since it sucked in version 5. Now it looks better
Looks better? But does it work better?
aaaaaabstraction heeeell
I am excite
create framework!
SSR is a scam to get you to pay for servers
its vit not vite???
Kotlin>
how much click bait on this channel ?
I always called it Vite, not Veet. Isn't it spelled Vite? 😂
Good video
Yeah, it's Vite for everyone I work with. Veet must be a US thing - that's a hair removal thing in the UK 😂
just open the first page of the Vite docs, it's literally the first sentence, they even have a recording of a proper pronunciation
and it's nothing new, they have it there since day one
It's from the french word, which stands for quick
@@bartek.igielski I was joking... 🙄
I never liked webpack, early user of snowpack, There is skypack but was a mess
Completely wrong about ESBUILD and why it gets used in vite
SSR is not the future
Not sure why people need all these tools. I've been building with HTML, CSS and vanilla JS in Notepad++ for the last 2 years, and never found a need for any frameworks etc. Admittedly, stuff takes longer but that's ok for me. I only land up with 3 files max to edit.
good for you, now you go to that corner while we talk real bidness
I'm early
long live webpack .. yall just bad
BRO, please make shorter interesting videos, I already work for 8 freaking hours to watch you read articles
you can't build no framework
No its not.
There's too many frameworks for basically doing the same thing, it's nice there are different options to choose from, but it just feels as if though people just keep creating new frameworks for the sake of making new frameworks around certain things, when one or two already exists and does most of or all of what they want already by copying the same features of the ones that already exist, but with little extra things added in that they want to see. Having too many different frameworks just complicates things, and if you ever try to step away from one to move to another, you end up having to change a lot of your stuff to convert to the new stuff.