The problem with this is that it changes the meaning of a kind 1 note. Users on clients that don't support this (all of them other than Amethyst) will always see the original note, leading to multiple versions existing *simultaneously*. Edit is a pandora's box.
Discussion
it's not but i think it needs to be done as replacement events without deleting old ones so people can see the full history
dealing with the diffs is a fools errand without a consensus so each update needs to be a full new version for this protocol, on the back end the storage can do the diffing to save space
i can see it working too... all events could be replaceable actually, if you just stop doing this racy delete events thing
i've thought about this very deeply for a very long time... diffs need consensus, you don't need to worry about that on the back end of a relay, it can diff the versions of an event, all you need is a "replaces" tag and put the event hash and the content replaces the event et voila
“Look, it says here you said it!”
“Nope! See!”
I agree it’s complicated but we should be careful about stopping innovation because of legacy data or apps. We’ve got what, a dozen widely used apps that display kind 1 events. We can do that upgrade. Users really want to be able to edit their own posts and i think collaborative editing opens up a whole new world, a new paradigm of social media.
I can see your point. The experience inside Amethyst is amazing, but when it doesn't replicate well outside on other clients there are some problems for sure.
I think it was a very innovative and interesting test. Very fresh, and still worthy some evaluation.
Clients don't need to repliate each other. This idea that we all need be exactly the same is terrible.
I agree with this!
I can also sense that there's some valid point in the fact that misunderstandings can come from notes looking different on different clients.
But indeed, at least each version is still signed by the authors actual nsec so it's still safe against malicious note insertions. Each note version is still authentically signed by the author.
I'll keep thinking about this for a while ^^
not to mention a relay attack vector