What ever happened to these discussions around storing files directly on relays (those who choose to accept file upload events)?

nostr:npub180cvv07tjdrrgpa0j7j7tmnyl2yr6yr7l8j4s3evf6u64th6gkwsyjh6w6 nostr:npub1xtscya34g58tk0z605fvr788k263gsu6cy9x0mhnm87echrgufzsevkk5s nostr:npub1gcxzte5zlkncx26j68ez60fzkvtkm9e0vrwdcvsjakxf9mu9qewqlfnj5z nostr:npub1acg6thl5psv62405rljzkj8spesceyfz2c32udakc2ak0dmvfeyse9p35c nostr:npub1l56z85xvaplh3033l8h2x9thypamu0d2hyqnxkw0ujuntqm8asaqh99sza nostr:npub12262qa4uhw7u8gdwlgmntqtv7aye8vdcmvszkqwgs0zchel6mz7s6cgrkj nostr:npub1skfmmcet2244yx8xn5xjfcmwg3u4mcjdndpd2gcjkq4qfycm6xtql0xa7d nostr:npub1jlrs53pkdfjnts29kveljul2sm0actt6n8dxrrzqcersttvcuv3qdjynqn

https://github.com/nostr-protocol/nips/pull/719

https://github.com/nostr-protocol/nips/pull/345

Reply to this note

Please Login to reply.

Discussion

Blossom

Websockets/json are not well suited to storing binary data. Blossom basically does this, but on http, and without asking relays to participate by default.

Horrible ideas die as fast as they get thought of

Can they be stored on several relays and then use the outbox model.

Blossom is better.

It works really well for files under 100KB. We are waiting for new Relay codebases to better manage large blobs. I think this is inevitable. But it doesn't work well on any of the current relay implementations.

One thing I remember hearing somebody point out is that the second you accept anything other than plain text you take on a whole new set of responsibilities & risks.

Not that I'm in any way knowledgeable enough to advise development, but I personally like the idea of relays being text only by default. Keeps things lightweight, and sure there can be bad things with text but the real bad stuff always needs to stay off Nostr infrastructure.

Very true. That’s why i think if this sort of thing is implemented, it should be in the form of specialized relays that accept file blobs, that way it can be an explicit decision of whether or not to accept those kinds of events.

The idea that binary payloads add risk is not correctly understood.

First of all, you can transfer virtually any binary payload via regular text events. You just need to encode it in text first. There are multiple ways to do this, but in practice, if you are afraid of binary payloads, you should be afraid of regular text payloads as well. There is no difference between the two. Criminals can use text encoding to transfer their binaries in plain sight. Legally speaking, the relay operator is as liable with text as with binary payloads. Pretending that this is not the case is not good legal defense.

And second, we already transfer many binary payloads. NIP-04 DMs are binary. Private zaps are binary. GiftWraps are binary. Private lists of people and bookmarks are binary. If binary itself was the issue for an operator, he/she shouldn't accept any of type of encryption, which will break significant usecases in Nostr, like private chats.

So, binaries by themselves shouldn't be the source of any extra risk .

The problem is more related to the software and hardware infrastructure each relay operator decides to run. I don't think there is going to be a silver bullet type of relay that solves it all and thus specialized operators will become more common as we move Nostr forward.