I was a happy Netlify user for years until they silently took down my app without warning. They blocked my account based on a bogus DMCA claim, the details of which point to a DIFFERENT app with a similar name. Unacceptable behavior from a hosting provider. I'll never use them again.
Oh, that's the best part. The support team sent me an incoherent message semi-explaining the ban and then blocked my email preventing me from responding to explain their mistake! Just stone-walled with no path for remediation. And just to be clear, I was not being rude/aggressive in my messages or anything like that.
As someone who has pissed away thousands of team hours working with the next.js abomination I’m very happy to see Astro take off.
I keep thinking “this looks like one of my very first terribly architected, organically created messes”. Astro may not be better, but just statistically speaking you probably aren’t going to be worse. Next is a house of cards.
The architecture is as bad as the code base. The only features anyone wants are perpetually in beta. The documentation is probably the only thing worse than the runtime.
We dropped SSR and used prerender.io to accomplish SEO and look back at that decision as a great one. I want to try Astro, looks more professional and propose built. Not at all interested in SaaS hosting.
Anything specific? We use Next very successfully, and, well it almost became the main way to use React nowadays... pretty good track record for an "abomination". It's not 100% there, I dislike some features built around Vercel offerings (eg. middleware) but it's... a good idea.
As an outsider: what exactly is it about "routers" that justifies so much change? Aren't there plenty of established designs to copy from other stateful UI frameworks?
I think supporting React Server Components is at the core of this issue. A ton of changes were necessary to support RSC and a different router is just one of many things that changed. Important features have been sacrificed to make it happen. You for instance no longer have access to the request/ response object in middleware or layouts. Currently playing with Remix as an alternative.
We are using Remix at work. We are super happy with it. None of the nonsense artificial limitations of Next, great integration with express, and a really simple model which, if you ever used a traditional “MVC” framework, will make you feel right at home.
Thought experiment: assume the worst: next.js + eleventy is in decline; what does it change in the short-term? It is likely that people using those frameworks are getting things done in them. To me, the fact this whole discussion is happening illustrates how fickle devs are in this particular ecosystem.
It's like nobody wants to be caught dead using something that might not be the new hotness.
The survey is a reflection of what our community responded.
Next is the largest framework in usage and is really will liked by it's users. This is clearly visible in the charts of the survey.
But for the first time we’ve run this survey, Next decreased in usage year over year (from 47% to 46%).
Astro jumped all the way from 11% to 18% with 87% responding they want to use it more.
Eleventy dropped in usage from our respondents from 19% last year to 16% this year.
None of this is an attack on anyone, it’s just data from our survey. The rise of Astro is one of the most newsworthy bits of data from the survey and reflects genuine excitement in the community we’re part of.
We changed the title to better reflect that the survey asks plenty of questions not directly related to jamstack (like usage of AI tools for web development etc).
Nothing changed in the methodology since last year and as always the survey was run by our data team.
If you scroll past the editorial part of what our team found newsworthy in the data and down to the actual survey data, you'll see clear charts of absolute framework usage and detailed breakdowns on satisfaction for each framework.
But the percentage changes are correct, aren't they? I don't think readers will think that the graphs represents absolute popularity, also (IMO) any survey taken by a company will be used for marketing purposes and should be consumed as such. (It's just some text on the internet, nobody is going to take it as gospel.)
It would be nice if the visualization reflected the information you've shared here (including both the absolutes and the amount of change). Then it wouldn't feel misleading and you would not need to write explainer comments on HN.
The absolutes are in a bar chart in the framework part of the survey that makes the absolute usage from the responses very clear. The chart Zack took issue with is from the editorial part of the survey where we found the biggest news in the framework section to be the growth and satisfaction for Astro.
This comes off as a disingenuous response, by typing out a large block of text while neglecting to address any of the individual concerns written in the article. "it’s just data from our survey" is a cop out -- the presentation of data is just as important as the data itself, and the concerns of underlying bias via communication model presented by Zach all seem reasonable to myself.
It's fine to not be an expert in Data Analysis, and if it's true that the company no longer employs someone with those skills, there should be a greater willingness to make adjustments and corrections when problems are raised with publicized analysis.
I really like the chart that he objects to. It is pretty clearly labeled as change and it helps highlight new and potentially interesting things. Would be neat to have a chart like that for lots of things—games on steam, news topics, community events, movies—anywhere you want to quickly grasp what the buzz is.
It doesn’t make next/eleventy look bad (of course they are ranked near the origin given their maturity and adoption). It makes Astro look like something worth investigating for a few minutes.
I don't have even a smidge of the background apparently required to make sense of this article. What is Netlify's battle with Vercel? What is their promotion and apparent now distancing from Jamstack about? Which frameworks would Netlify have ulterior motives to promote or disparage?
This article is written for the kind of person who reads articles on the topics like state of web development in 2023.
Vercel is big enough and important enough that such a reader will undoubtedly know of it. That reader will probably, but not certainly, also know about eleventy.
Netlify, who arguably started or at least encouraged the space Vercel inhabits, is understandably pissed that they aren't in Vercel's shoes today. Eleventy is pissed that they are receiving a bit of collateral damage in that messaging war.
Somewhat an outsider (but I have used Netlify and eleventy). It's basically a tale as old as time: trying to build a sustainable business around open source technology and perhaps fleeting trends in software development.
Vercel and Netlify offer essentially the same thing: easy hosting of mostly static websites + a few bells and whistles. They each support development of frameworks (eleventy, Next.js, Gatsby, and perhaps others) that make their hosting business more compelling to use. Netlify branded it's approach as "Jamstack" which I believe stood for "Javascript and some other Marketing Stuff". Now that the winds of change are blowing, Netlify want to reposition themselves as something a bit different and down play the old stuff and also not make it look like their rivals are doing to well.
Oh and some of the React core team landed at Vercel, and React has been focusing on backend features and partnering with Next/Vercel on early development of said features.
So there is frustration that the React project is now perhaps an extension of Vercel's interests, and focused on helping them sell hosting.
It was my (perhaps outdated) experience that in fact some elements of JS (like using scraper libraries) would not work on Vercel at the time due to memory of serverless function and other limits. But locally they worked on my dev machine.
When my team cut our monthly Netlify costs by $1.5k, Netlify didn't inquire why, contrasting with GCP's prompt response to a similar spending reduction. Our departure from Netlify was due to significant bugs in Netlify's Next.js Edge and double billing us when Git contributors didn't map cleanly to Netlify accounts, despite us using Github OAuth. Support individuals were great, but it seems like their advocacy for us fell on deaf ears within the Product/Engineering camp.
Even more puzzling was that a few months prior, their partnerships team had been working with us to get our customers' ecommerce storefronts on the platform. Netlify's (then) head of partnerships left and nobody else followed up ever. Our customers are now pointed at Fly.io instead, and everyone is quite happy with that recommendation.
My anecdote, and this article, seems like evidence of Netlify's reactivity to Vercel over customer-oriented growth. Vercel, in contrast, is more focused on developing open-source projects that organically brings people in.
I remember being astonished at the time of Netlify's rise that someone had the audacity to compete with (a slice of) Amazon S3, and heartened to see what looked like success (usage). Alas, it seems it was a zero interest mirage, and the situation is as it always was, that only the bigs have sustainable business models in web services, especially the more open, generic ones.
edit: what's with the downvotes, did I offend someone?
I wish them well, but the success indicators I've seen from them have been similar; funding rounds and users. They are certainly better positioned, since running hot nodejs servers is a higher revenue proposition than white-labeling blob storage. Question is, does control of the Next.js project constitute a big enough of a moat to keep profitable customers from shifting their hosting to big clouds. The decline of Heroku, for example, does not make me confident.
