←back to thread

Local-First Software Is Easier to Scale

(elijahpotter.dev)
175 points chilipepperhott | 1 comments | 05 Jul 25 15:54 UTC | HN request time: 0s | source
Show context
myflash13 ◴[05 Jul 25 17:30 UTC] No.44474202[source]▶
Closest thing to true “serverless”: entire MVC app (Django/Rails/Laravel) in the browser with WASM and data persistence by SQLite over CDN.

All the server has to do then is serve binaries, all the business logic is in the client.

replies(2): >>44474217 #>>44474669 #
gjsman-1000 ◴[05 Jul 25 17:31 UTC] No.44474217[source]▶
Brilliant… but now you need to validate that the client did all their business logic correctly without tampering. That alone can be so complex it defeats the point.
replies(1): >>44474340 #
justinrubek ◴[05 Jul 25 17:49 UTC] No.44474340[source]▶
No... you don't need that. Not for the overwhelmingly vast majority of cases. Let people use their own software. Tampering? Not my problem. Let people do it if they want.
replies(3): >>44474357 #>>44474599 #>>44474690 #
gjsman-1000 ◴[05 Jul 25 17:52 UTC] No.44474357[source]▶
Anything that runs as a SaaS, or B2B, has that issue… which is the overwhelming majority of software.

Anything that requires sharing information with other users is also a pain in the neck, as you basically need to treat your internal logic like a proprietary, potentially hostile, file format.

replies(1): >>44474456 #
1. Fire-Dragon-DoL ◴[05 Jul 25 18:10 UTC] No.44474456[source]▶
There is a lot of SaaS that is essentially "for the buyer to the buyer", what I mean is that the software doesn't provide content to somebody else, or there is no incentive to serve malicious content (e. g. B2B). Why would tampering be relevant in those cases?

There are situations where it's relevant, but I don't think it's as many as you say