nicholasoxford/fastlyDream

Upgrade to fastify-vite 2.1.3

Opened this issue · 3 comments

Hey, this looks great, but you might want to set the fastify-vite version to ^2.1.3 -- previous 2.x releases were broken :(

Also, let me know if you need any assistance, I'll be glad to help.

Hey, @galvez, how has your experiencing been deploying? In some ways I am tempted to do vercel + serverless, but I was really loving fastify!!

For more clarification, I want the absolute fastest site (don't we all). Where would you deploy to achieve that?

Fastify can have above average performance on Lambda with aws-lambda-fastify. I think the very same module can be used for deploying fastify-vite applications to Vercel with ease. I haven't tried it yet, but I'm pretty sure it's doable and it's been proven to have awesome performance stats.

As for myself, personally, I'm interested in building the next Vercel and Netlify of sorts, so to speak. But it won't be a massive commercial juggernaut. I'm interested in building an open source solution users can 1-click-deploy to a $60/mo dedicated machine from Linode (or Hetzner, or similar) and just host hundreds of sites on their own at a fraction of the cost.

That is interesting. Personally for me the app.get() syntax was mindBlown when I got it to work. If I could easily deploy both the front end and server.js in a decoupled, asynchronous manner good shit. I would also like to one click change from pointing to my localhost server.js to a hosted one. I also really like vercel secrets CLI functionality for adding environment variables.

I love the idea of one server to host them all. If it was as easy to host a nuxt site, as a Vite ssr, or vue3 + fastify that would be great

So a little bit about this site, I am really interested in computer generated art. I've done style transfer stuff, but computer generated art from scratch (zero-shot) is just insane to me. On my macbook it took like ten hours to produce an image from big-sleep on my local. Using a Nvidia K80 took like an hour and 50 minutes, but using AWS p3.xlarge (Tesla V100 GPUs) it takes 189 seconds. So I have a bunch of hosted art at www.dreamland.to, but I have completely rewritten the front end and a bunch of AWS work for you to pay 5 bucks, feed it an input, and get an image sent back in 5 minutes to your email. I am so close to being done