refs:

• avoid custom core-eval for contract deployment #9717
• feat: deploy contract without custom code (WIP) dapp-agoric-basics#48
• how does a user get powers for use with contractStarter? agoric-labs/ag-power-tools#14
• how to limit excess authority when registering a storage node #8594
• how to grant limited authority to save contract upgrade materials? #8378

Description

Drive most of the core-eval code from ContractMeta and permit.

• generic Issuer / Brand handling
  • issuerKeywordRecord: derive from permit.issuer.consume
  • issuer / brand publishing: drive from permit.issuer.produce
  • extra BootstrapPowers: derive from meta.name, permit.issuer.produce, permit.brand.produce
• meta.adminRoles generalizes sending oracle invitations
• ContractMeta: move to a new .meta file so both contract and core-eval can get it without swallowing the other whole
  • likewise: FastUSDCConfigShape -> meta.deployConfigShape
  • contractName = meta.name
• permit: move to fast-usdc.meta file to separate it from orch.start code
  • factor out typical permit stuff: orchPermit,
• makePrivateArgs() remains custom code, though only for stuff beyond orchestrationPowers
• finishDeploy() remains custom code: for publishFeedPolicy and calling creatorFacet methods

permit is down to ~15 LOC:

In order to keep it type-safe, meta and permit are imported. Maybe there's a way to parameterize the types or something so that the custom code can call the boilerplate. hm.

DRAFT until:

• rebase on master (I started this while reviewing test(multichain): fusdc "advance and settle" as a macro #10692)
• rename fast-usdc.start.js to orch.start.js and update builders, boot tests

Security Considerations

This contributes somewhat to...

• how to limit excess authority when registering a storage node #8594
• how to grant limited authority to save contract upgrade materials? #8378

Scaling Considerations

n/a

Documentation Considerations

should simplify Writing a Core Eval Script to Deploy a Contract

Testing Considerations

For Fast USDC, existing tests should do it. It's essentially a pure refactor, but some of the filenames have runtime impact in builders and such, so it's possible that the changes are visible from tests.

Upgrade Considerations

This could / should make it less likely to mess up saving admin facets.