Next js gets so much hate on HN. In my experience, it's been great. It's kind of like Apple in that, if you buy into the ecosystem, you unlock great benefits. Fwiw we're running on cloudflare at my work and it's been one of the most enjoyable DX I've had in a long time. RTFM and you'll be fine. Listen to conference talks and you'll understand why some of the 'weird' decisions were made - they might not even seem so weird anymore.
Tangential, does any heavyweight framework allow user to hit back and it loads exact content and scroll position, out of the box? HTMX does it perfectly, since its "unit of serialization" is html dom instead of components and internal state.
On one end you have something like https://vike.dev, very light framework. It might not have all the bells and whistles like Next but your code is going to be your code and if you change your mind you should be able to switch to anything in the future.
In the middle you have something like Astro or Remix. More creature comforts but they try to be close to web standards. Remix forms can work without js.
Then on the opposite end you have Next. Monkey patching fetch method, headache inducing cache behavior (now probably finally gone by default), common breaking changes, strange middleware implementation, relying on unstable React features and most importantly the only-on-Vercel features: Incremental Static Regeneration, image optimization etc. No other framework needs sth like https://opennext.js.org/
Thank you. I've seen 'only-on-Vercel features' mentioned in the past, but so far, I have yet to find any specific examples of classic 'vendor-lock' in a sense 'if you use this it will not work anywhere except for Vercel'. AFAIK you can self-host NextJS apps with docker, albeit some features may not be as efficient as on Vercel they will work.
This seems to be the case for both of the examples mentioned—ISR [1] and image optimization can be done on CloudFlare for example. Am I missing something?
Don't get me wrong I'm not affiliated with Vercel, I'd like to think we can still switch from Vercel-hosted NextJS to self-hosted Docker in the future if we want which was my impression so far.
Yes luckily a lot of talented people are sharing their workarounds so you can host on AWS or Cloudflare with most feature parity and in plain nodejs docker with some loss. But it's a constant game of catch-up for these folks.
We have some marketing sites that are static and Next on cloudflare is working perfectly for us at a large company. Fair enough if you don't like it. It is the most popular framework (or top 3) and hasn't scared off its user-base in spite of the issues you listed. Maybe those aren't the primary concerns (or dealbreakers) of most projects?
Marketing sites are usually short-lived. They can die with the platform and it's fine.
The OP is building their product on top of Next and was asking about lock-in. I work with React codebases that are 8+ years old. Luckily no-one chose sth like Next so there is a flurry of additional libs, some fell out of fashion today but none of them prevented the app to evolve or be deployed wherever.
There is so much money in Vercel that I don't expect Next to die anytime soon but if sth more cool appears people on Next will have hard time migrating. Where someone sticking to more vanilla React can even easily switch to Solid or another non-React jsx solution.
Another example: Flutter has 12k issues, but that's because it's the most popular cross-platform app framework. Also it's not all literally issues, it's just a tracking board for all kinds of "things" that need to be tracked.
(I work on Next) The issue count includes core issues, examples, and docs. We might consider breaking that out in the future. Next has ~1.2M active devs, so we expect a certain level of issues being opened weekly that will take time to investigate. With that being said, we definitely still want to get the total # down (and have many old issues that are stale and don't have repros we need to clean up)
Next js gets so much hate on HN. In my experience, it's been great. It's kind of like Apple in that, if you buy into the ecosystem, you unlock great benefits. Fwiw we're running on cloudflare at my work and it's been one of the most enjoyable DX I've had in a long time. RTFM and you'll be fine. Listen to conference talks and you'll understand why some of the 'weird' decisions were made - they might not even seem so weird anymore.
Tangential, does any heavyweight framework allow user to hit back and it loads exact content and scroll position, out of the box? HTMX does it perfectly, since its "unit of serialization" is html dom instead of components and internal state.
All I can see is more lock in and less interoperability with anything else. Seems like a liability to be honest
Hi, we are building our product in NextJS, and I was wondering what you meant by lock-in and less interoperability. Thanks!
On one end you have something like https://vike.dev, very light framework. It might not have all the bells and whistles like Next but your code is going to be your code and if you change your mind you should be able to switch to anything in the future.
In the middle you have something like Astro or Remix. More creature comforts but they try to be close to web standards. Remix forms can work without js.
Then on the opposite end you have Next. Monkey patching fetch method, headache inducing cache behavior (now probably finally gone by default), common breaking changes, strange middleware implementation, relying on unstable React features and most importantly the only-on-Vercel features: Incremental Static Regeneration, image optimization etc. No other framework needs sth like https://opennext.js.org/
Thank you. I've seen 'only-on-Vercel features' mentioned in the past, but so far, I have yet to find any specific examples of classic 'vendor-lock' in a sense 'if you use this it will not work anywhere except for Vercel'. AFAIK you can self-host NextJS apps with docker, albeit some features may not be as efficient as on Vercel they will work. This seems to be the case for both of the examples mentioned—ISR [1] and image optimization can be done on CloudFlare for example. Am I missing something? Don't get me wrong I'm not affiliated with Vercel, I'd like to think we can still switch from Vercel-hosted NextJS to self-hosted Docker in the future if we want which was my impression so far.
[1] https://stackoverflow.com/questions/75526234/how-can-i-use-n...
This is covered in the linked blog post for Next 15, including a demo repo, videos, and new templates here: https://github.com/nextjs
Yes luckily a lot of talented people are sharing their workarounds so you can host on AWS or Cloudflare with most feature parity and in plain nodejs docker with some loss. But it's a constant game of catch-up for these folks.
We have some marketing sites that are static and Next on cloudflare is working perfectly for us at a large company. Fair enough if you don't like it. It is the most popular framework (or top 3) and hasn't scared off its user-base in spite of the issues you listed. Maybe those aren't the primary concerns (or dealbreakers) of most projects?
Marketing sites are usually short-lived. They can die with the platform and it's fine.
The OP is building their product on top of Next and was asking about lock-in. I work with React codebases that are 8+ years old. Luckily no-one chose sth like Next so there is a flurry of additional libs, some fell out of fashion today but none of them prevented the app to evolve or be deployed wherever.
There is so much money in Vercel that I don't expect Next to die anytime soon but if sth more cool appears people on Next will have hard time migrating. Where someone sticking to more vanilla React can even easily switch to Solid or another non-React jsx solution.
Great release! People should be much happier with the new caching defaults.
“We've focused heavily on stability”
Fantastic, it only has 2550 open bugs now! /s
Without any frame of reference this seems to simply be a function of the popularity and usage of a complex project like this.
Chromium has 120 new issues opened so far today. Does that therefore make it a bad browser engine doomed to failure?
This.
Another example: Flutter has 12k issues, but that's because it's the most popular cross-platform app framework. Also it's not all literally issues, it's just a tracking board for all kinds of "things" that need to be tracked.
(I work on Next) The issue count includes core issues, examples, and docs. We might consider breaking that out in the future. Next has ~1.2M active devs, so we expect a certain level of issues being opened weekly that will take time to investigate. With that being said, we definitely still want to get the total # down (and have many old issues that are stale and don't have repros we need to clean up)
Fair, I’ll provide some reference.
Next has 130k stars and 2550 issues labeled as bugs.
Svelte has 80k stars and 64 issues labeled bugs.
Vue has 47k stars and 80 issues labels bugs.
Remix has 30k stars and 240 issues labeled bugs.
HTMX has 38k stars and 27 issues labeled as bugs.