should we store everything in just one ipfs file ?

[virgile-dev]

at the moment we are sending a file to ipfs per transaction. might be more efficient to put everything in a single file that gets updated (hash would change everytime).

Might not be that accountable as it won't be as easy to retrieve info as we would break the rule 1 transaction = 1 input data ipfs hash

[AlexJupiter]

Sat with @LucasIsasmendi now and we don't think this is possible due to performance problems.

Our understanding is that if you need just one user (e.g. for a specific ballot) you would need to query all users and then filter down to that specific individual.

Proposing we close this?

[aecc]

@AlexJupiter however in the current approach we are querying ipfs for every user data. Did you discussed/concluded which is more performant: 1 query to IPFS + filtering user data or N queries to IPFS?
As IPFS storage is basically free we could maybe even think about an hybrid approach, where we store 1 ipfs hash per user and an ipfs hash with all users

[AlexJupiter]

Did you discussed/concluded which is more performant: 1 query to IPFS + filtering user data or N queries to IPFS?

When @LucasIsasmendi explained it to me, the N queries definitely sounded more performant. The thought of querying loads of users, just to get one, seemed silly to me (however this is outside of my expertise).

As IPFS storage is basically free we could maybe even think about an hybrid approach, where we store 1 ipfs hash per user and an ipfs hash with all users

Why not then!

[LucasIsasmendi]

The issue with 1 file to store all the content is a performance issue related with UX, imagine if we wait to get all the tweets before start to explore them