😐
computers can't lie
Discussion
Rust-analyzer lies nonstop about compile errors. I don't know why it's so finicky. Starting to understand why the graybeards prefer terminal editors.
code definitely abstracts the terminal env for the various compilers/LSPs.
which project is a pain today?
Hashpool, same as everyday. I'm having to go off script to manually implement serialization and deserialization. I'm not sure if it's a good idea to allow variable sized keysets. Might be simpler to just fix the number of keys and stick with the boilerplate fixed size Sv2 encodings. I don't know if this will cause problems later on down the road. Thoughts?
Start with fixed. Don't rabbit hole on making things variable when a constant will suffic to land an MVP.
iirc, we were going to use the keyset to impl vardiff-like with the target hash. I would think that a keyset size of 8 would be sufficient to provide weighting for shares.
Yeah I hear you. I am trying to encode what cdk gives me. Not even sure if I can or should customize the keyset right now. That could be a different rabbit hole.
Not even sure if keyset is needed. Maybe we can use one pubkey per epoch and let the share difficulty speak for itself? This will have implications when splitting or combining ehash tokens. We are straying from the cashu spec now, which may be inevitable but do we want to rush into it?
It's a tradeoff of customizing cdk to suit hashpool needs or customizing Sv2 to suit cdk. At this point, I don't know the answer. Need to try a bunch of shit and feel my way around the problem to get a good grasp on it.
Recreate the data strives in Sv2 then create Into and From traits between the two structs