timdaub on Warpcast

David Furlong pfp

Long time coming, but going to properly shut down discove.xyz Jan 1st; any remaining users, export your data/feed algo code that you want to keep. It costs too much to run & maintain, given nr of custom feeds it caches and given its hosted on vercel. Might relaunch at some point/could open source if enough interest

23 replies

15 recasts

53 reactions

timdaub pfp

what are the monthly costs?

1 reply

0 recast

0 reaction

David Furlong pfp

4 replies

1 recast

2 reactions

David Furlong pfp

/memes turns out a million slow requests (to cache feeds) a day is a lot

3 replies

0 recast

2 reactions

timdaub pfp

kiwi still 25€ a month

2 replies

0 recast

0 reaction

David Furlong pfp

yeah very different infra need lol

1 reply

0 recast

1 reaction

timdaub pfp

I also have global edge caching but for free with Cloudflare. I have everything a next js has

1 reply

0 recast

0 reaction

David Furlong pfp

lol thats a different kind of caching

1 reply

0 recast

0 reaction

timdaub pfp

1 reply

0 recast

0 reaction

David Furlong pfp

discove: - custom feeds can take up to 2 mins to run on SQL, can't let first user going to that feed in that time period get an uncached version, or they wait for 2 mins for the page to load. That's what just cloudflare edge cache would do - cache is stale (expires) in less than 5 mins (as chrono/algo social feeds are)

1 reply

0 recast

1 reaction

Dan Romero pfp

almost looks like you're subsidizing inefficient sql written by no-coders to the tune of 2k a month. that's 👀 about how easy it is to write expensive queries

2 replies

0 recast

1 reaction

timdaub pfp

I think this is still too charitable because operation costs are now a factor to discontinue an otherwise well-working product that has potential

1 reply

0 recast

0 reaction

David Furlong pfp

its also unmaintained lol - still relies on warpcast apis

2 replies

0 recast

0 reaction

David Furlong pfp

doesn’t support eth usernames etc. very broken and I’d rather put my energy now into mod and supporting other apps on Farcaster

2 replies

0 recast

1 reaction

timdaub pfp

that‘s fair but keeping in running is like a out of the money option. Maybe it works like shit but not selling the options is not yet realizing a loss. And markets change all the time

1 reply

0 recast

0 reaction

timdaub pfp

I actually think u should spend the time to build it back to keep it sustainable and running. I bet you will win later in some way bc of that. But I would try to make very high bang buck refactors to make it sustainable at the expense of removing most features maybe

1 reply

0 recast

0 reaction

timdaub pfp

eg I keep rugpullindex.com without core functionality online just bc it benefits me credibility wise

1 reply

0 recast

0 reaction

timdaub pfp

@df if u disable frequent caching would you save 1-2k a month and the site would technically still work?

1 reply

0 recast

0 reaction

David Furlong pfp

would be slow as hell and still buggy bc unmaintained. also warpcast apis die soon and I tried to migrate off but new database schema is even harder to index well and still a lot of work to get working

1 reply

0 recast

0 reaction

timdaub pfp

- slow is better than non existent - buggy too - Warpcast APIs die when they die. That isn‘t in your control and you are speculating. You can still shut down at zero cost when they shut them down. - don‘t migrate don‘t fix, just control short term operating costs

2 replies

0 recast

0 reaction

David Furlong pfp

tim im gonna trust my product judgement here

1 reply

0 recast

2 reactions

David Furlong pfp

like i said if enough ppl want it I would consider open sourcing, but itll take some hrs work to make sure its scrubbed of any api keys and similar that may have been committed

1 reply

0 recast

1 reaction