ethersphere / swarm

Swarm: Censorship resistant storage and communication infrastructure for a truly sovereign digital society

https://swarm.ethereum.org/

GNU Lesser General Public License v3.0

488 stars 112 forks source link

sw3 contracts meta + timeline #559

Open zelig opened 6 years ago

zelig commented 6 years ago

after hackdays 2018-12-05 in Budapest:

Swap

[ ] rewrite Swap tests so they don't affect each other (almost done already)
[ ] moving cheque timeouts into cheques
[ ] remove hardcoded timeouts in other places (e.g. hard depo decreases)
[ ] define a format for the channel deposit allocation table
[ ] implement the necessary change for soft deposits (CADT submission + refutation)
[ ] bounties in Swap
[ ] completing the Swap tests
[ ] updating the docs (a bit outdated now)

Swear / Swindle

[ ] rewrite witness interface and trial rules to work with role assignments
[ ] rewrite Swindle with roles
[ ] rewrite Swear contracts in a way so anyone can start cases and the commitment is presented in the first trial stage
[ ] implement a sample trial for mailbox services to see if it makes sense

zelig commented 5 years ago

Obsolete tasklist created as a comment.

README (https://github.com/ethersphere/swap-swear-and-swindle/README.md)

Ralph Pichler @ralph-pichler May 16 10:46

here are some things that need experimentation if somebody wants to work on the contracts:

witness:

better witness interface
- often note_id is just some custom payload (see OracleTrial) (the remark = sha3(rules, sha3(payload data)) hack sucks)
- maybe keep using remark for swap-initiated swear deposits but forward the payload to the witnesses (not the note_id, if the witness really cares about the note it can still be included in the payload)

swap:

add custom timeouts for hard deposits (critical for swear integration)
mayble multiple timeouts for the same beneficiary (ideally with changes offchain?)
unify cheques and notes logic (implement index for notes, when does replacment make sense?)
- always pass fields as arguments to avoid contract storage

swear:

think about how this works with (potentially continous) payment on successful service
clean up the messy commitment code (many fields are probably not needed)
make the plaintiff related logic a bit more sane (right now anybody can start a trial for every commitment and potentially get the deposit)
support taking only part of the deposit in the offchain case (swap witness would have to receive payout amount)

trials:

create more test trials and witnesses for potential real world applications to see wether they can work with the witness / state machine concept

let me know if somebody wants to tackle something in this list so we avoid too much duplicate work