Hacker News Reader

Comment by EGreg | Hacker News Reader

Sorry but how is it serverless? Is it a static HTML file with JS that does the job? Isn't there a server that the request goes to?

View on HN

Good question. It depends on your definition of serverless. Strictly speaking there is no serverless, there's always some infrastructure somewhere.

However the calendar file gets created out of thin air from the GET parameters alone. There is no state or storage involved. Other calendar apps have a backend with a DB; this one doesn't. Now whether that counts as truly serverless is up to you I guess. It's as close as you could get with a calendar I'd say. Not easy to get the point across in a title. ;) Hope that helps.

serverless is when you save the state but don't keep a machine sitting around when there are no requests, so you'd keep your calendar state in a hosted db and when someone asks for the calendar, a thread spins up and creates the calendar from the state and then disappears until there is another request. This is basically the exact opposite of serverless

This can easily run on Cloudflare workers, you don't even need a server

... what do you think those are

this thread is making me feel like I'm going crazy

Mmm yes, this one reminds me of the microservices debacle from a few years ago about when does your monolith technically become a microservice. I can't find the link, but I think we were unable to clearly define at which point your monolith has become a true 'SoA' group of microservices.

The generous interpretation of what op said is that you don’t have to maintain a server in any way or shape which is indeed the case. You only write business logic while the infrastructure automatically connects it to handle inbound requests.

> serverless is when you [...] don't keep a machine sitting around when there are no requests

FTFY

The key part of serverless is that you don't have a server idling around. Whether or not you use a hosted db is not relevant.

And you may even use some 'serverless' DB services. Or not. The app is still serverless.

"Stateless" is probably clearer. "Serverless" usually refers to services from cloud providers where your code is run on demand rather than having a server that is idling around (to use a sibling comment's phrase). I don't really care for "serverless" as a term because there are still servers (as you point out) and there are other architectures and cloud services that seem equally serverless that aren't usually called that.

Your app is always running to answer those GET so it's not serverless. Serverless means that your rust code would be invoked upon GET.

Isn't it serverless when it's local-only? Or, in an extreme case, peer-to-peer?

Reply on news.ycombinator.com