He also hopped off it and said so in this video. What's your point? How many people who watch Theo even are going to adopt into Edge anyways? It's okay to try things out.
@@LeonBlade Try things out is this: "Let me try this, then I'll come back to you with some results, but be aware, this is just a try/error thing. Don't do it at home.". Another thing is being paid/sponsored to advocate, to stand up for something and then, later, come back here and say "Oh, I know why it doesn't work. Of course, here is the reason on this video". There's no problem in being wrong. First words of his video should be: "I was wrong. I am sorry. Now I will tell you why I was wrong and now I'm going to try to give you some advice and, hopefully, it can help you out". And that is not a subtle thing..
@@LeonBlade I think the issue is that. this guy has no "weight" but just flies around in some half-baked conceptualization of software development. Being a coding influencer rather than an actual system developer these days?
I love how Vercel always walk back whatever paradigm they pushed a year ago, just before announcing a brand new rendering paradigm that's definitely manna from heaven, this time!
Yeah, it's great that they try out new things, but really if you are building something on a larger scale you better not use something that didn't have the test of time yet. Being an early adopter can give a good feeling, but in this industry many times you might end up with your tail between your legs. Let others find out the blockers for a brand new framework/package/service, and if the hype still doesn't die down after 3 or 4 years you can pretty safely pick it up. Also there's this bias about new technologies - for one technology that stood the test of time ten has been buried in the graveyard. We might not hear about these technologies, only about the success of the single one and assume that everything that is hyped will be a massive success and is ready to be used.
@@ParaNormanable There's exactly 0 wrong with edge. There's a lot wrong with 'trivial edge without data replication'. Yes, it is unsuitable for cluetards or people trying to do trivial things simply. Yes, you're going to have to write your own code as opposed to glorified declarative specification of pre-defined behaviours provided by some behemoth framework, and you're going to need clue. The big problem realistically is that it is largely JS-centric.
We are currently working on containerized micro frontends, where your site will be split up into different browser sections and each new section is a randomly generated javascript framework written by AI. On each request you will have a new JS framework running your things. We think it's the most JS-like thing ever and we think it will take over the web!
The edge thing will only work for companies like Cloudflare and Fastly that owns their own infrastructure and can offer a much cheaper pricing by compacting the edge v8 runtimes and get CPU-time pricing instead of runtime pricing. It won't scale for Vercel which is just running AWS Lambda and have the same cost model as original nodejs based runtime.
Vercel isn't just running AWS Lambda. They only do that for their Serverless offering. For the Edge offering they have their own Edge Runtime which is a set of APIs that interface with V8 isolates (likely run on Cloudflare Workers).
Indeed, web technology has never been easier, and it's intrinsically Client/Server based, meaning it works on the edge by default. Theo seems to constantly bash client rendering as if it's a bad thing. Speed is not a good reason for SSR, the only thing I would say it potentially gains is SEO. But saying that any sites I've done client side SPA seem to index fine on all major search engines, so not even sure SEO is a good reason. ps. Also even that 200-300ms delay can be reduced if it's just for First Contentful Paint (FCP) reasons, just bootstrap the HTML with some inline styled content, very similar to how mobile apps have a splash screen.
@@dinoabdurahmanovic 200ms is huge. If you've never worked on a large consumer-facing site you won't know this, but adding 200ms to a page load is going to lose you a massive amount of traffic, whether it's from SERPs or PPC. In the PPC case, this can cost you on the order of thousands of dollars every hour (obviously scales with PPC budget).
@@maciekdeveloper even if that is true, vercel was also responsible for testing that technology and others in huge production environments, and thanks to it we now have an actual valid and tested conclusion of what is better, therefore we had progress (thanks to vercel)
Na, old folk who know how things work knew it was just fad hype like most of the web technologies turn out to be. The old 'let's do this differently because it's cool' usually turns out to just be a VC cash grab. Don't fall for it again.
Haha classic, if you use Vercel (and therefore NextJS) for an app that needs to live for several years, if you want to maintain and keep it current, you will have to rewrite it 10x over (at least) to accommodate all their endless experiments and paradigm shifts which have no real benefit to your business and all aimed to lock you deeper into their ecosystem. All the freebies and templates and silver-platter starter projects are just bait for their crap and they can't even keep those up to date with the pace of their changes.
I never understand these sorts of takes, do you think other ecosystems have kind of “stung” us into always feeling like we need the latest version of everything? It’s perfectly fine to keep using old Vercel APIs if you dont have resources to rewrite your stuff all the time.
More like investors are knocking, Vercel is trying to reduce their server expenditure. Shutting down the least used, and a lot expensive product to begin with.
I'm glad I didn't jump on this "edge" or even "serverless" nonsense because of hype. I mean I also didn't need it on apps that I've worked on. I invested my time getting to know Linux more and understanding how to deploy things on a VPS/VM on my own without these managed services and I can say that I learned a ton. Yeah it's great to know these managed services, but I think knowing the barebones is a more valuable skill long term.
Not me rewriting my NextAuth jwt code because I am forced by the library to make it compatible with Edge(even though i don't use it), just so Vercel gives up on Edge today 😂 Unlucky timing, but at least I learned something new i guess
As someone who really leveled up during this whole edge craze its been an absolute blast nerding out on the edge runtimes, how theyre built, performance, quirks and drawbacks and building around them. Its nice to see we arent just doing things to do them on edge anymore and are taking a measured approach to building with them and finding good improvements.
@@marcuss.abildskov7175 Sure for production stuff workers are awesome, I am still allowed to love the rust forks of it and other specialty stuff for AWS etc.
6 หลายเดือนก่อน +1
I'm done with all this. From now on, everything in the same place, no new hyped stuff, backend is near the DB, VPS, Docker, auth near. I will not care if it takes more time ... I built a web app 13 years ago this way, it still works. I could not launch a web app that I built with this new stuff changing the logic every 2-3 months. I am done, by by...
Man, that's why I love your videos. Yeah, you talked about news but you always have a deep understanding on the subjects and you bring us to a deep dive in it. I can't say how much I learned just watching your videos ! Amazing job, mate ;)
Doesn't TH-cam also use PPR? Or at least something similar. When you open TH-cam it basically immediately comes up with a page skeleton that then bit by bit gets filled with the dynamic content.
Next "NEXTJS gave up on server components", back to render the UI at our super powerful smartphones ! Nahh, that won't happen because they need to sell us the server side rendering
I had never considered building my side projects on any of these edge solutions because of the database rtt problem. Lately I've been building an app using turso as the database and it occurs to me that turso's embedded replicas perfectly solve that problem.
Called it from the outset. All of the edge stuff is a cancerous outgrowth of the microservices hype era. The fastest experience for users will always be a properly configured and optimized majestic monolith (or regionalized set) with all resources necessary to compute a response in one place, and a CDN to serve pre-rendered sections and their static assets. The fact that anyone thought it was a good idea to spin up an entire virtual machine and init relevant JS libraries to serve a request (aka cold start) is beyond me. People forget that the V8 engine powering node may start slower, but that's only because it optimizes code more (and better) which makes it more performant over time (and use less memory!) than JSC and other Edge runtimes.
I still dont understand how far people went in this trip, now they say we learn something new, well everyone who knows basic infra already knew it would not work, you don't need to spend millions into something a lot of people knew it would fail. monoliths are easier to deal with and most of the time faster and cost efficient, but some frontend-only devs want to move React to the server so they can work in backend, but while they do it, they recreate the wheel multiple times creating their own hell
Isn’t that partial pre-rendering strategy the same as Astro’s islands? Astro does not ship js to the browser by default, if one wants to add interactive, one creates “islands” of interactivity and that is the only js shipped.
Yeah, Astro's approach is really suited to the vast majority of websites out there. You can then pick making your interactivity client-rendered or SSR. Then if SSR, using a node server or edge deployment. Each is an architectural decision to be made depending on the website. Going so hard on edge was always a curious choice.
That's why people seem to like ThePrimeagen more. He sounds a lot more informative as opposed to Theo who sounds a lot more authorative. Although both of them share useful information. ThePrimeagen is obviously the better human being though.
I think I'm going to just stick with doing things the "old fashioned" way. Vite to build my frontend, deployed on CDN for quick loading to users browsers. Keep it a SPA, and stream in data over websockets so we get first bytes quickly, and extra data gets cached into localstorage. Can make this super snappy and also support offline usage (happens more often than people realize, spotty cell coverage, airplanes, etc). The recent allergic reaction to running JS on client computers is crazy to me. Sure, we can massively reduce the bloat (Vite with Svelte or similar can solve that largely), but just because some folks ship too much JS doesn't mean we have to go crazy and compute everything on the server-side like it's 1999 and we are still writing PHP. It seems too convenient to me that all these companies are pushing us to move compute back to servers, when that means paying them more per month! Conflict of interest. So I don't trust anything Vercel says, or any of these other companies that are doing similar (Deno Deploy, etc).
Code split + CDN, it would get super snappy app, this idea of rendering SSR looks to me like a regretion, I came from Rails and the main point for me to learn React was to reduce the computing load from the server to the client, its crazy to me to ask to go back to SSR, especially now that client devices are faster, why would anyone compute rendering stuff in the server lol, also you lose the offline support.
PPR solves an already solved issue. You just need to accept that not everything needs to be put inside react. Why not serve the static part of the site directly, use react for the main content stuff and use some other thing to glue them together? Why must everything live inside the react monolith?
@@hamza_dev Your app in SSR also needs the client to make a request to the server far away from the client and I do hope its close to the database otherwise it will be super slow
@@masadamsahid Most apps don't need to be after you pass the login page and yes most of the crawler can handle SPA nowdays this is not a really good reason to use SSR.
For those in the comments: the context which allows this to exist, already existed. They just change its behavior a little bit and save quite an amount of time, 300ms is a tangible amount. Up to 50% faster responses, if what they said is true.
what if instead of using NodeJS you're just writing your website in a language that compiles it into an efficient binary that doesn't need a complex runtime?
Didn't jump on the edge train. Didn't make sense from a complexity stand point. Now it's broken down and in the rear view mirror. Hype trains are dangerous to board.
Isn't this just the same as astro or solid island architecture but having to tune config files to play nicely with vercel? And along the same lines. Wouldn't Qwik's resumability just be an inherently better and less copy-able "hydration" than this partial prerendering? I wouldn't claim to be super knowledgeable on all these, so I'd like criticism/support on this topic, but it just seems to me that astro/solid is already leaps ahead of next with qwik being in another lane that has potential to be bounds ahead of all the others.
I prefer my 'Shitty office chair'!. It cost LESS than a gamers chair, and is arguably more comfortable and longer lasting. I've had my 2 office chairs since 2015, and they are still just as good as the day I purchased them, however my gamers chair that I never use, is already falling to pieces, and it cost MORE than 1 of my office chairs. I swear, put the word 'gamer' on it, charge 2 times more than a regular office one, ???, profit! The stupid fad needs to end.
Maybe, I agree, not really a fan of all those stupid super colorful chairs, some with RGB and trinkets, seriously? On the other hand, you can find some ridiculous headmaster CEO chief of bullshit chairs, compared to them, RGB chief of circus chairs become tolerable. I have a simple non expensive chair made in china, no colors and specific "gaming" trinkets chair, it's actually quite good, comfy and practical, don't care for the gaming part of those chairs, and I play games.
@@StupidusMaximusTheFirst Mmmm, unicorn vomit. I actively avoided it when I built my current computer back in April-2023. It wasn't easy though, I hate that RGB is opt-out when it should be opt-in.
Why do you hate calling BS on a sponsor? I think saying stuff like that makes me be cautious about ur takes in the future, eventho you did call them out on a minor thing just now, maybe just weird wording idk
so we can make nextjs middleware not-trash now, right? "sorry you cant use any part of the node library in middleware because one day you might want to host on vercel and use edge" has to be among the worst parts of next. Also worth noting; aws is already working on solving the cold start problem with their own runtime. so that part of the problem will go away without vercel having to invest in doing it themselves.
Just want to say I had a wrong idea about the quality of your content for some reason. I’m glad to be wrong. Thanks for sharing this, it was very informative
Theo pushed for edge so hard. Remember this kids
grain of salt
He also hopped off it and said so in this video. What's your point? How many people who watch Theo even are going to adopt into Edge anyways? It's okay to try things out.
@@LeonBlade Try things out is this: "Let me try this, then I'll come back to you with some results, but be aware, this is just a try/error thing. Don't do it at home.". Another thing is being paid/sponsored to advocate, to stand up for something and then, later, come back here and say "Oh, I know why it doesn't work. Of course, here is the reason on this video". There's no problem in being wrong. First words of his video should be: "I was wrong. I am sorry. Now I will tell you why I was wrong and now I'm going to try to give you some advice and, hopefully, it can help you out". And that is not a subtle thing..
@@LeonBlade I think the issue is that. this guy has no "weight" but just flies around in some half-baked conceptualization of software development. Being a coding influencer rather than an actual system developer these days?
The edging stream where Theo took a break to blow.
my main goal is to explode
i edged to this video
what is the edge
My man out here rocking that 80's CIA agent undercover as a cocaine runner for the Colombians look, and pulling it the fuck off.
Prequel movie to Monk featuring a young Stottlemeyer
Too prim for that. More Studio 56 if you ask me
Bro,its Colombians, from Colombia, not like the film studio. 😁
@@SlavojZizekEspanol-libros FIXED AND PUSHED TO MAIN.
You heard it here first, vercel is now in it's goon phase.
lmfaoooooooooooooooo
real
😂😂
That's VC funding for ya
All this time and effort reinventing client-server, ajax, and cdns. This is a 25 year old workflow.
Old is the new new. 🤣😂🤣😂
My thoughts exactly. I haven't changed much about what i've been doing for the last two decades and I've been fine.
Agreed, I'm sitting here and scratching my head thinking "is HTML streaming something special or is just ajax?" 😂
i couldn't stop thinking about framesets with the PPR-Thing and how the world looked like if we just made them better and useable.
still waiting for when we stop using js on the server... 5-10 more years?
I love how Vercel always walk back whatever paradigm they pushed a year ago, just before announcing a brand new rendering paradigm that's definitely manna from heaven, this time!
Yeah, it's great that they try out new things, but really if you are building something on a larger scale you better not use something that didn't have the test of time yet. Being an early adopter can give a good feeling, but in this industry many times you might end up with your tail between your legs. Let others find out the blockers for a brand new framework/package/service, and if the hype still doesn't die down after 3 or 4 years you can pretty safely pick it up. Also there's this bias about new technologies - for one technology that stood the test of time ten has been buried in the graveyard. We might not hear about these technologies, only about the success of the single one and assume that everything that is hyped will be a massive success and is ready to be used.
@@ParaNormanable There's exactly 0 wrong with edge. There's a lot wrong with 'trivial edge without data replication'. Yes, it is unsuitable for cluetards or people trying to do trivial things simply. Yes, you're going to have to write your own code as opposed to glorified declarative specification of pre-defined behaviours provided by some behemoth framework, and you're going to need clue. The big problem realistically is that it is largely JS-centric.
Welcome to VC funding
Edge is not in thousands of locations. The largest edge network is cloudflare which is in around 300+ locations
Cloudflare is the master of edging
@@DylanJava your mom taught you well, bravo
Edge is literally in the thousands of locations when you're talking about IoT.
@@marcuss.abildskov7175 but we are not talking about IoT. We are talking about vercel
I’m gonna wait another five years and then see again if frontend has agreed with each other
At that point they'll have found another thing to disagree on, so you'll be waiting forever 😅.
in my 10 yrs of experience, Frontend never agrees on anything
We are currently working on containerized micro frontends, where your site will be split up into different browser sections and each new section is a randomly generated javascript framework written by AI. On each request you will have a new JS framework running your things. We think it's the most JS-like thing ever and we think it will take over the web!
Love the content but seeing your face making some weird expression on the cover of every single video just makes it so hard to want to click on it.
you should blame the viewers instead for not clicking on vids without silly thumbnails and making this a necessity
Unfortunately, if creators don't use thumbnails like this, they don't get the same reach. It's stupid (imo) but necessary.
Yeah we just need the face not the bad expressions 😂
seeing such crybabies in comment section of every single video just makes me 🤢
I just came here to say that but damn man you took word out my mouth
Never stop edging
The edge thing will only work for companies like Cloudflare and Fastly that owns their own infrastructure and can offer a much cheaper pricing by compacting the edge v8 runtimes and get CPU-time pricing instead of runtime pricing. It won't scale for Vercel which is just running AWS Lambda and have the same cost model as original nodejs based runtime.
Vercel isn't just running AWS Lambda. They only do that for their Serverless offering. For the Edge offering they have their own Edge Runtime which is a set of APIs that interface with V8 isolates (likely run on Cloudflare Workers).
I thought this was about Edge, the browser.
if only
So did I
I thought this was about Edge, the wrestler
edge the sport?
Everyone has given up on that Edge already lol
Australia mentioned let’s go
It amazes to me that we are adding this amount of complexity to save 200-300ms, many of these ideas are the opposite of good engineering.
I totally agree.
Over-engineering to save like 200ms or 20kb.
This is on each request, when hitting the like btn and what not. For most apps absolutely unnecessary.
Indeed, web technology has never been easier, and it's intrinsically Client/Server based, meaning it works on the edge by default. Theo seems to constantly bash client rendering as if it's a bad thing. Speed is not a good reason for SSR, the only thing I would say it potentially gains is SEO. But saying that any sites I've done client side SPA seem to index fine on all major search engines, so not even sure SEO is a good reason. ps. Also even that 200-300ms delay can be reduced if it's just for First Contentful Paint (FCP) reasons, just bootstrap the HTML with some inline styled content, very similar to how mobile apps have a splash screen.
If you're trying to hit SLAs then 200ms is a godsend
@@dinoabdurahmanovic 200ms is huge. If you've never worked on a large consumer-facing site you won't know this, but adding 200ms to a page load is going to lose you a massive amount of traffic, whether it's from SERPs or PPC. In the PPC case, this can cost you on the order of thousands of dollars every hour (obviously scales with PPC budget).
This field is humbling. Every now and then we have to appreciate the fact that we didn't know and accept reality.
not really, only idiots fell for edge shilling
Vercel made money and returned to node 😂😂
@@maciekdeveloper even if that is true, vercel was also responsible for testing that technology and others in huge production environments, and thanks to it we now have an actual valid and tested conclusion of what is better, therefore we had progress (thanks to vercel)
@@upsxace You're right
Na, old folk who know how things work knew it was just fad hype like most of the web technologies turn out to be.
The old 'let's do this differently because it's cool' usually turns out to just be a VC cash grab. Don't fall for it again.
Haha classic, if you use Vercel (and therefore NextJS) for an app that needs to live for several years, if you want to maintain and keep it current, you will have to rewrite it 10x over (at least) to accommodate all their endless experiments and paradigm shifts which have no real benefit to your business and all aimed to lock you deeper into their ecosystem. All the freebies and templates and silver-platter starter projects are just bait for their crap and they can't even keep those up to date with the pace of their changes.
They're not deprecating anything?
I never understand these sorts of takes, do you think other ecosystems have kind of “stung” us into always feeling like we need the latest version of everything? It’s perfectly fine to keep using old Vercel APIs if you dont have resources to rewrite your stuff all the time.
Normally people would praise a framework that manages to simultaneously push the boundaries of performance and maintain backwards compatibility.
What is this mindless hate? This comment have nothing to do
@@Blu3yo They probably don't write any code, they just write TH-cam comments
More like investors are knocking, Vercel is trying to reduce their server expenditure. Shutting down the least used, and a lot expensive product to begin with.
I'm glad I didn't jump on this "edge" or even "serverless" nonsense because of hype. I mean I also didn't need it on apps that I've worked on.
I invested my time getting to know Linux more and understanding how to deploy things on a VPS/VM on my own without these managed services and I can say that I learned a ton. Yeah it's great to know these managed services, but I think knowing the barebones is a more valuable skill long term.
I saw this with no context thinking we were talking about Microsoft Edge for like the first minute or so
Good on Vercel for admitting something didn't work out and pivoting.
next month "NEXTJS gave up on server components". the javascript ecosystem is a joke
Nahh, that won't happen because they need to sell us the server side rendering face-green-smiling
server rendering is amazing, chill with the nonsense hate
How is edge thing JS ecosystem?
Not me rewriting my NextAuth jwt code because I am forced by the library to make it compatible with Edge(even though i don't use it), just so Vercel gives up on Edge today 😂
Unlucky timing, but at least I learned something new i guess
As someone who really leveled up during this whole edge craze its been an absolute blast nerding out on the edge runtimes, how theyre built, performance, quirks and drawbacks and building around them. Its nice to see we arent just doing things to do them on edge anymore and are taking a measured approach to building with them and finding good improvements.
What edge runtimes? The only one that's worth looking into is the one Cloudflare Workers is using.
@@marcuss.abildskov7175 Sure for production stuff workers are awesome, I am still allowed to love the rust forks of it and other specialty stuff for AWS etc.
I'm done with all this. From now on, everything in the same place, no new hyped stuff, backend is near the DB, VPS, Docker, auth near. I will not care if it takes more time ... I built a web app 13 years ago this way, it still works. I could not launch a web app that I built with this new stuff changing the logic every 2-3 months. I am done, by by...
Theres potential for a meme here, with the thumbnail faces and edging and all that
i dont care how vercel feels, im going to keep on edging
a perfect example of not having FOMO about new tech stuff - it will die sooner than you'll have a chance to test it out :)
The reality is that most of the apps you will work on are made With React 16-17 and Redux with (redux-saga/redux-thunks/graphql), its what its
Man, that's why I love your videos. Yeah, you talked about news but you always have a deep understanding on the subjects and you bring us to a deep dive in it. I can't say how much I learned just watching your videos ! Amazing job, mate ;)
Doesn't TH-cam also use PPR? Or at least something similar. When you open TH-cam it basically immediately comes up with a page skeleton that then bit by bit gets filled with the dynamic content.
Next "NEXTJS gave up on server components", back to render the UI at our super powerful smartphones !
Nahh, that won't happen because they need to sell us the server side rendering
100% true
You had me at ”This isn’t clickbait”
I knew this would happen and when I saw this title I shouted "Called it!" out loud
I REALLY WANTED TO EDGE WITH THIS ONE😥😥😥🔥🔥🔥🗣🗣🗣
I had never considered building my side projects on any of these edge solutions because of the database rtt problem. Lately I've been building an app using turso as the database and it occurs to me that turso's embedded replicas perfectly solve that problem.
Called it from the outset. All of the edge stuff is a cancerous outgrowth of the microservices hype era.
The fastest experience for users will always be a properly configured and optimized majestic monolith (or regionalized set) with all resources necessary to compute a response in one place, and a CDN to serve pre-rendered sections and their static assets.
The fact that anyone thought it was a good idea to spin up an entire virtual machine and init relevant JS libraries to serve a request (aka cold start) is beyond me.
People forget that the V8 engine powering node may start slower, but that's only because it optimizes code more (and better) which makes it more performant over time (and use less memory!) than JSC and other Edge runtimes.
I still dont understand how far people went in this trip, now they say we learn something new, well everyone who knows basic infra already knew it would not work, you don't need to spend millions into something a lot of people knew it would fail. monoliths are easier to deal with and most of the time faster and cost efficient, but some frontend-only devs want to move React to the server so they can work in backend, but while they do it, they recreate the wheel multiple times creating their own hell
crab is for Rust
oh, I thought it was a disease they caught
edging to this rn
Guess they have been "Edge" ing for a while. only to never see the "climax" of their work.
Isn’t that partial pre-rendering strategy the same as Astro’s islands? Astro does not ship js to the browser by default, if one wants to add interactive, one creates “islands” of interactivity and that is the only js shipped.
Yeah, Astro's approach is really suited to the vast majority of websites out there. You can then pick making your interactivity client-rendered or SSR. Then if SSR, using a node server or edge deployment. Each is an architectural decision to be made depending on the website. Going so hard on edge was always a curious choice.
First time discovering this channel and I've always liked NextJs
"if u don't give me a better name you're stupid" -- dawg chill out lmao
Yeah, Theo's streams are toxic af
That's why people seem to like ThePrimeagen more. He sounds a lot more informative as opposed to Theo who sounds a lot more authorative. Although both of them share useful information. ThePrimeagen is obviously the better human being though.
I think I'm going to just stick with doing things the "old fashioned" way. Vite to build my frontend, deployed on CDN for quick loading to users browsers. Keep it a SPA, and stream in data over websockets so we get first bytes quickly, and extra data gets cached into localstorage. Can make this super snappy and also support offline usage (happens more often than people realize, spotty cell coverage, airplanes, etc). The recent allergic reaction to running JS on client computers is crazy to me. Sure, we can massively reduce the bloat (Vite with Svelte or similar can solve that largely), but just because some folks ship too much JS doesn't mean we have to go crazy and compute everything on the server-side like it's 1999 and we are still writing PHP. It seems too convenient to me that all these companies are pushing us to move compute back to servers, when that means paying them more per month! Conflict of interest. So I don't trust anything Vercel says, or any of these other companies that are doing similar (Deno Deploy, etc).
Code split + CDN, it would get super snappy app, this idea of rendering SSR looks to me like a regretion, I came from Rails and the main point for me to learn React was to reduce the computing load from the server to the client, its crazy to me to ask to go back to SSR, especially now that client devices are faster, why would anyone compute rendering stuff in the server lol, also you lose the offline support.
PPR solves an already solved issue. You just need to accept that not everything needs to be put inside react. Why not serve the static part of the site directly, use react for the main content stuff and use some other thing to glue them together? Why must everything live inside the react monolith?
Post the link to what you're reading in the description please :/
"None of these are running node"
AWS: Hold my lambdas
Vercel’s done with edging?
We can still carry on edging though
SPA + code splitting = done ... no need for complex infrastructure
When you are working a global scale that doesn't make sense anymore.
Your SPA still needs to fetch the data from a server.
Will that still SEO-friendly?
@@hamza_dev Your app in SSR also needs the client to make a request to the server far away from the client and I do hope its close to the database otherwise it will be super slow
@@masadamsahid Most apps don't need to be after you pass the login page and yes most of the crawler can handle SPA nowdays this is not a really good reason to use SSR.
This really is the year of serverlessness
How long till they update middleware’s so they can run on node runtime?
When will someone tell Theo that Remix has had the ability to do all of these things since day 1 😭😭😭😭😭
why they dont use bun for their backend?
working fine in my Microsoft edge..
Is Vercel still pushing the jamstack architecture or have they abandoned that too?
1:52 when I saw Edge with capital E in the title, I thought it was about the browser.
edging is old fashion, everyone is gooning rn
Please link to original articles and tweets in the description.
another day, another switch in js
@@HUEHUEUHEPony i am that htmx guy :)
At first I thought the Edge browser, then I thought CDN. Never heard of the Edge runtime before.
What's the drawing app you use?
For those in the comments: the context which allows this to exist, already existed. They just change its behavior a little bit and save quite an amount of time, 300ms is a tangible amount. Up to 50% faster responses, if what they said is true.
so nothing changes for next.js devs?
what if instead of using NodeJS you're just writing your website in a language that compiles it into an efficient binary that doesn't need a complex runtime?
I didn’t quite understood, could you show us how you edge?
would have really appreciated a link in the description to the tweet in question
This video really took the edge off.
anyone that did not get the obvious - seems like them functions infra will be rewritten in rust. Like the aws's llrt
what is that canvas/drawing tool?
Didn't jump on the edge train. Didn't make sense from a complexity stand point. Now it's broken down and in the rear view mirror.
Hype trains are dangerous to board.
Doesn’t the quik framework stream everything, literally.
with partial pre rendering, can you revalidate (ISR) the data for the static parts of the page that lives on the edge?
I think edge runtime is stupid, you should call it edging time
I am on the edge with Vercel.
jk, love them!
Great video. Thanks for the heads up I’m in the middle of launching an app with “edge” built in functionality may have to do some refactoring 🤔
They abandoned their child
EdgingTime
PERFECT Name
there was a firebase solution
Isn't this just the same as astro or solid island architecture but having to tune config files to play nicely with vercel? And along the same lines. Wouldn't Qwik's resumability just be an inherently better and less copy-able "hydration" than this partial prerendering?
I wouldn't claim to be super knowledgeable on all these, so I'd like criticism/support on this topic, but it just seems to me that astro/solid is already leaps ahead of next with qwik being in another lane that has potential to be bounds ahead of all the others.
vercel pivoting to gooning
Sad to see Vercel losing their edge streak
I prefer my 'Shitty office chair'!.
It cost LESS than a gamers chair, and is arguably more comfortable and longer lasting. I've had my 2 office chairs since 2015, and they are still just as good as the day I purchased them, however my gamers chair that I never use, is already falling to pieces, and it cost MORE than 1 of my office chairs.
I swear, put the word 'gamer' on it, charge 2 times more than a regular office one, ???, profit! The stupid fad needs to end.
To be fair my Herman Miller definitely costs more than any gaming chair lol
Maybe, I agree, not really a fan of all those stupid super colorful chairs, some with RGB and trinkets, seriously? On the other hand, you can find some ridiculous headmaster CEO chief of bullshit chairs, compared to them, RGB chief of circus chairs become tolerable. I have a simple non expensive chair made in china, no colors and specific "gaming" trinkets chair, it's actually quite good, comfy and practical, don't care for the gaming part of those chairs, and I play games.
@@StupidusMaximusTheFirst Mmmm, unicorn vomit. I actively avoided it when I built my current computer back in April-2023.
It wasn't easy though, I hate that RGB is opt-out when it should be opt-in.
@@StupidusMaximusTheFirstchief of circus lol
Why do you hate calling BS on a sponsor? I think saying stuff like that makes me be cautious about ur takes in the future, eventho you did call them out on a minor thing just now, maybe just weird wording idk
What does SF pilled mean?
Name for the edge runtime? - Edge Runtime
sir theo can you make a video for honojs and elysiajs
What is the diagram software being used?
Finally! Considering returning to next
edge runtime == v8 isolate
I thought this was somehow about the Microsoft Edge Browser xD
No more edging?
I'm quite confused by all this edging
so vercel is gonna get rusty? ok
node 20 being fast is kind of weird XD
I wonder how it compares to bun
anyone know what “SF Pilled” means?
so we can make nextjs middleware not-trash now, right? "sorry you cant use any part of the node library in middleware because one day you might want to host on vercel and use edge" has to be among the worst parts of next.
Also worth noting; aws is already working on solving the cold start problem with their own runtime. so that part of the problem will go away without vercel having to invest in doing it themselves.
I also wonder what cratb 🦀 could be ? 🤔
Just want to say I had a wrong idea about the quality of your content for some reason. I’m glad to be wrong. Thanks for sharing this, it was very informative
I thought it was something related to the browser...
where is the tweet link?
After so much complexity and so many bugs... No wonder why
Oh, Sorry, I kinda thought this was about browser compatibility issues.
2:20 isnt this just cdns
Wait... what? Seriously? I'm just learning edge. wt... dude! Okay, gonna listen to the video now....