Nuh 🔻 on Nostr: I don't want to speak for the team, but since this design was mine I will take the ...
I don't want to speak for the team, but since this design was mine I will take the "blame" ... yes the indexer returns the data.
The reasoning here is that while it could just give you a URL, that is not going to be enough for this app requirements, namely;
1. Data availability and latency shouldn't be subject to the variance of homeservers availability and latency.
2. Some transformations are needed, for example inlining rich text things like showing the ID as the current name of the user in a mention, or for example compressing images.
3. Some data, in fact the most important data can't be deduced from the homeserver at all, for example backlinks and replies and counts of tags etc.
So honestly you have to trust the indexer to get the quality and features the app required from us.
There are room for improvement, most importantly falling back to the homeserver if the indexer is forced to censoring something or someone.
I want to note here that if the client was signing the data before sending it to the homeserver, none of these limitations would go away, which is one of the reasons I didn't invest in client side signing and the key management that that entails.
The protocol (client-homeserver)can get more sovereign and more private and more encrypted... but that won't change the nature of pubky app.
If the first app was private chat, i would have made different choices, but these choices weren't worth the time investment if at rhe end you have to trust the indexer to serve you things that can't be signed anyways.
The reasoning here is that while it could just give you a URL, that is not going to be enough for this app requirements, namely;
1. Data availability and latency shouldn't be subject to the variance of homeservers availability and latency.
2. Some transformations are needed, for example inlining rich text things like showing the ID as the current name of the user in a mention, or for example compressing images.
3. Some data, in fact the most important data can't be deduced from the homeserver at all, for example backlinks and replies and counts of tags etc.
So honestly you have to trust the indexer to get the quality and features the app required from us.
There are room for improvement, most importantly falling back to the homeserver if the indexer is forced to censoring something or someone.
I want to note here that if the client was signing the data before sending it to the homeserver, none of these limitations would go away, which is one of the reasons I didn't invest in client side signing and the key management that that entails.
The protocol (client-homeserver)can get more sovereign and more private and more encrypted... but that won't change the nature of pubky app.
If the first app was private chat, i would have made different choices, but these choices weren't worth the time investment if at rhe end you have to trust the indexer to serve you things that can't be signed anyways.