Has every client stopped offering an option to use imgproxy for autoresizing images?
I remember users could configure their own imgproxy on Snort, Iris and Coracle. Was that a bad idea because no one was using the option and if anyone used they would have a sitting duck imgproxy that anyone else could abuse?
We could think of some solutions for community-hosted imgproxy servers that can be secure by having the client compute the required HMACs locally, or is this too ambitious?
@hodlbod @utxo the webmaster 🧑💻 @npub1v0lx...qj49 @Vitor Pamplona @Sirius @npub1syjm...f6wl
Login to reply
Replies (14)
Tengo una pregunta mejor: ¿por qué preocuparse por imagenes cuando el precio de Bitcoin llega a 69.387 dólares? ¿No es hora de dejar de perder tiempo con detalles y enfocarse en lo importante?
Its still enabled by default in Snort
@The Fishcake (nostr.build) convinced me it was bad from an optimization standpoint. I would like to see this stuff built into blossom but I'm not exactly sure how.
Or ipfs. Also distributed
I have been using IPFS for a while and I do feel that it could help for more decentralized storage, but seeing what happened with the latest updates, I think that network still needs more development, because with only one update that was made to the file cube, many of the nodes stopped responding to previous nodes and trying to change everything you stored from a node with an old version to a newer one is not very intuitive.
bruh
😂
why is it bad?
A personal image proxy that you could configure would be neat. I would use that to avoid the issues of centralized image proxies
I can't remember why it was bad. I think part of it had to do with the fact that nostr.build optimizes images itself and he wanted clients to use those directly so that they didn't have to serve the full sized file over and over. But caching on the proxy's side should improve that, and clients don't know to download the smaller files anyway.
and not all images are hosted on nostr build anyways, actually most are not these days
And we wouldn't want them to be
I think (and I don’t recall exactly) it was mainly due to caching of potential CSAM and such at that layer. We try to catch and delete them quickly but if you cache it’ll stay. Of course if you perform scanning and reviewing yourself for that cache, that is fine. And I think the deletion was an issue too, where caches usually do not respect TTL that hosts serves.
What are the issues of centralized image proxies?