A place for developers to talk about building on Farcaster.

/dev

while all the warpcast infra crumbles (and 3rd party clients too) it is perfect timing to remind my infra philosophy

keep infra simple, move as much computing as possible on client, vertically scale backend as much as possible before going horizontal

blog.borodutch.com/i-never-use-ssr-and-you-shouldnt-either/

cto @ bwl.gg (lunchbreak.com), 80M+ users (bdut.ch) 🙏 built @remindme, @mintit, @gpt, @healthbot, farcantasy.xyz

SSR does not mean “lambdas on vercel”, which i agree, yeah $$$. I deploy remix anywhere that runs node, for example.

Read heavy workloads like blogs are always easy to cache at the edges, CSR/SSR whatever. 

Most services fall over because of i/o not cpu, no "rendering modes" are immune to that.

memebuilding • prev manifold.xyz, twilio.com

“static frontends without backend” are not worth discussing, it’s trivial to scale.

If it’s dynamic at all, like feeds, then you’ll prob be io bound (yes which leads to cpu/oom). Rendering mode is largely irrelevant. Same data reqs.

I’ve handled millions rps just fine. So have many others.

how to say you never handled 10m of requests per second without saying you never handled 10m of requests per second

static frontends without backend served over cdn are immune to i/o issues

an no, most services that get popular *do* fail because of cpu/ram limits