@mauve my first thought alsooo
@fleeky @mauve okay but how do you do serverless with p2p? because isn't that actually a bit of a holy grail? take any public cloud's serverless API and reimplement it backed by #Veilid, #I2P, #Holochain? which one has everything you need to reimplement literally any single major commercial serverless platform?
really hoping someone's got a quippy response spring loaded here 🤞🏻
@travisfw @jeremy_list @fleeky I think that's the direction the IPVM foks were going. https://github.com/ipvm-wg/homestar/
In P2P apps what you can do is move that "server code" onto the client directly. Then you focus on local first DBs and sync engines for getting clients on the same page.
The big question with "serverless functions" is how you can manage authority and trust. Apps that require secrets to access a DB need trusted servers. At that point you either need creds per client or skip central DBs
@travisfw @jeremy_list @fleeky "scaling up can mean blowing your budget if you don't configure limits" is the thing that goes away when you stop using centralized DBs and infra and move data / logic to peers. If data gets popular it has more peers resharing it to the network instead of costing individuals more. As a dev you pay nothing (but can offer backup services), as a user the devs can't take your info/community/utility from you so long as the p2p network lives.
@travisfw @jeremy_list @fleeky Another direction is using p2p to bridge to self hosting where you host servers (or worker runners) on community or personal infra that you trust, and the p2p connections (like veilid) skip the need for setting up SSL/public IPs to talk to compute from client apps.
In the case of AWS who is paying to keep the infra running? If it's not the clients themselves then you end up with crypoto coin economics like holochain which is less appealing than stuff being free