This is an artificial PR designed to focus reviewer attention on specific changes. Merger or non-merger is irrelevant; and branch asz/transaction-consensus will continue to be the most up-to-date topic branch.
This PR captures a sequence of commits following a dependency update. The new dependency on tasm-lib master points to triton-vm v42-alpha6, which in turn points to the most up-to-date version of twenty-first, which drops the field leaf_index from MmrMembershipProof. This last elision is what required a lot of attention to a) make compile and b) fix consequently failing tests.
A lot of the work to anticipate this change was already done, but you really only know the impact of a change once things start breaking down.
Here is what I want the reviewer's attention on: a lot of the failing tests deal with mutator sets somehow. The tests pass now, but I'm not sure I fixed them in the right way. Divining their intention was difficult, the mechanics of batch applying records with batches of to-be-preserved membership proofs is notoriously difficult. I'm afraid I a) neutered tests in otherwise benign ways; or worse b) neutered tests while breaking the (up till now) (probably) correct mutator set logic.
This is an artificial PR designed to focus reviewer attention on specific changes. Merger or non-merger is irrelevant; and branch
asz/transaction-consensuswill continue to be the most up-to-date topic branch.This PR captures a sequence of commits following a dependency update. The new dependency on
tasm-libmaster points totriton-vmv42-alpha6, which in turn points to the most up-to-date version oftwenty-first, which drops the fieldleaf_indexfromMmrMembershipProof. This last elision is what required a lot of attention to a) make compile and b) fix consequently failing tests.A lot of the work to anticipate this change was already done, but you really only know the impact of a change once things start breaking down.
Here is what I want the reviewer's attention on: a lot of the failing tests deal with mutator sets somehow. The tests pass now, but I'm not sure I fixed them in the right way. Divining their intention was difficult, the mechanics of batch applying records with batches of to-be-preserved membership proofs is notoriously difficult. I'm afraid I a) neutered tests in otherwise benign ways; or worse b) neutered tests while breaking the (up till now) (probably) correct mutator set logic.