Farcaster Devs

If Warpcast was open source, what are 1-2 things you would want to learn how we do?

frames-devs

So… I may or may not have stumped @horsefacts.eth and nerd-sniped @cassie into discovering a race condition in Warpcast’s OpenGraph scraper with my *wildly innovative* approach to rendering a SvelteKit website inside of a frame 🙈

Resolution and caching of frame and image URLs.

(Warpcast still can’t seem to consistently decide if skiffle.dev is a frame or an image 😕)

Resolution and caching of frame and image URLs.

(Warpcast still can’t seem to consistently decide if skiffle.dev is a frame or an image 😕)

https://warpcast.com/darrylyeo/0xbca1753e

/frontend dev・founder/decentralized data wrangler /blockhead・UI/UX abstractionist・DeFi/web3 builder・dropout・24・prev /dydx, ritual.net・x.com/darryl__yeo

SKIFFLE overloads URLs to return either frame HTML or an image based on HTTP content-type negotiation. Hasn’t worked as expected and it's hard to debug without knowing how the resolver works. Specifically:

1) the order it checks URLs for images or frame data

2) what Accept headers are sent when

3) caching behavior

Two things I think would help:

1) If Warpcast knows a URL is meant to be a frame image based on the meta tags, it should send an `Accept` header equivalent to the HTML `<img>` tag in browsers e.g. `image/*`

2) Cache should be indexed by both URL and embed type, not just URL (assuming that’s how it currently works)

FarCon

BDFL of Quilibrium, Eng @ Farcaster, ex-Coinbase, always opinionated, never hydrated

/quilibrium

cc @cassie (we chatted about this during /farcon)