Consolidate shared functions for stateless and manager - #175
Merged
Conversation
The stateless splice tests ran one contract shape: an enum contract over a single oracle, spliced by the party that offered it. The manager path covers both shapes, several oracles, thresholds below the oracle count and either party initiating. A splice builds a new contract, so the shape of what it produces is what the CET selection and the adaptor signatures then work on, and none of that was exercised. The stateless splice section is now one driver. It takes a contract shape, the rounds to run — who offers each replacement, and which way the collateral moves — and the party that settles the last one. Every contract in the chain is built from that shape over an event of its own, so each round settles on its own attestation. `ContractShape` and `ShapedContract` in stateless_utils carry the shape: enum, numeric, numeric that tolerates an oracle spread, and disjoint, each over any oracle count and threshold. A `ShapedContract` keeps the oracles it announced with, so the scenario that funded the contract attests the event afterwards without tracking them. That takes the stateless splice tests from 6 to 19: enum over one, three and five oracles, numeric over one and three, numeric with difference over five, a disjoint contract settled on either of its two events, accept-party splices at multi-oracle and numeric shapes, either party settling, and four chains. The manager splice path had two holes of its own. It always settled from the party that offered the first contract, so the accepting party never broadcast the CET of a spliced contract. And accept-party splices only ran on an enum contract over one oracle. `TestPath::Splice` now carries a `SplicePath` that names the settling party, and seven tests cover the rest: accept-party splices on enum 3-of-5, numerical with difference 3-of-5 and a disjoint contract, accept-party settlement on enum 3-of-3 and numerical 3-of-3, an accept-party splice closed by hand, and the two existing chains now settle from the accepting side. 15 of these ran against a live regtest chain here, including every new shape and both settling parties.
Both suites built the same things by hand: oracles announcing an enum event over the same four outcomes and a digit decomposition event over the same base and digit count, an enum descriptor that pays alternating parties, a numeric descriptor rounded to the satoshi, and the contract those descriptors go into. What kept them apart is the layer each hands the contract to. `ddk-manager` takes a `ContractInput`, which names oracles by public key and event id. `ddk::contract` takes the wire `ContractInfo`, which carries the announcements themselves. `ddk_testenv::dlc::ContractLeg` builds either form from one descriptor: an announcement already carries the oracle public key and the event id, which is all the manager form needs. The new module holds the constants, the oracle mechanics — announce, sign, collect attestations — both descriptor builders, the two contract forms, and `SpliceDelta`. It sits behind the `dlc` feature of `ddk-testenv`, already a dev-dependency of both crates and of nothing else. That feature depends on `ddk`, which depends on `ddk-manager`, both of which depend on `ddk-testenv` in turn: a cycle cargo allows because theirs are dev-dependencies. Policy stays with each suite, because they differ on purpose. The manager tests pick a random outcome and a random oracle subset above the threshold; the stateless tests settle on a fixed outcome with the first `threshold` oracles. The payout curves differ too — the manager tests build polynomial and hyperbola pieces, the stateless tests take a straight line from `ddk-payouts` — so `numeric_descriptor` takes the curve instead of choosing one. One behaviour is kept deliberately. `attest_numeric_event` offers every candidate outcome to every oracle and lets the first one stand, because an oracle that already signed an event refuses the rest. The shared signing functions ignore that refusal for the same reason, and `attestations` asserts an oracle really signed before its attestation is used. The two harnesses drop 481 lines and add back 209 that call into the 436 line module they now share. Verified on a live regtest chain after the move: 7 manager execution paths (enum, disjoint with and without difference, numerical 3-of-3, single funded, refund, and a numerical splice), 4 stateless ones (enum, numeric with difference 3-of-5, disjoint settled on its numeric event, and a numeric splice), and the 31 offline stateless tests.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.