fix(zcf): mint kind during reincarnation #7508
Conversation
| /** @type {import('@agoric/vat-data').Baggage} */ zcfBaggage, | ||
| /** @type {{ (keyword: string, issuerRecord: IssuerRecord): void }} */ recordIssuer, | ||
| /** @type {GetAssetKindByBrand} */ getAssetKindByBrand, | ||
| /** @type { (exit?: undefined) => { zcfSeat: any; userSeat: Promise<UserSeat> }} */ makeEmptySeatKit, | ||
| /** @type {ZcfMintReallocator} */ reallocator, |
There was a problem hiding this comment.
This seems like a radical shift in style from @param. Why? At least at first, I do not find it more readable.
There was a problem hiding this comment.
It's the style that "infer params from usage" automates. I think it's worth allowing because it removes the need to repeat the parameter name.
| /** @type {string} */ keyword, | ||
| /** @type {IssuerRecord} */ issuerRecord, |
| ) => { | ||
| /** @type {SetStore<ZCFMint>} The set of baggages for zcfMints */ | ||
| /** @type {SetStore<import('@agoric/vat-data').Baggage>} The set of baggages for zcfMints */ |
There was a problem hiding this comment.
Is there any way to make this type more specific, to declare that it holds a set of zcfMintBaggage (which has a particular set of keys), rather than a set of generic Baggage (which itself is just a MapStore)?
There was a problem hiding this comment.
It would be pretty awkward since a MapStore allows any keys. We could have a subtype that maintains a list of keyname to type mappings and conditions get/init/set accordingly but I'm not sure it's a good idea and it's definitely out of scope :)
| zcfMintBaggageSet.add(zcfMint); | ||
| return zcfMint; | ||
|
|
||
| zcfMintBaggage.init('issuerRecord', issuerRecord); |
There was a problem hiding this comment.
This works, but it now looks awfully like per-zcfMint state, and I wouldn't mind a cleanup pass that replaced the prepareSingleton (one Kind per mint) with a prepare something else that shared a single Kind among all mints, with state.keyword, state.zoeMint, and state.issuerRecord. That would make startup faster, remove the extra Kinds, reduce memory use a little bit, and be generally simpler.
But, probably out of scope for this PR.
There was a problem hiding this comment.
Good idea. I'll follow up with that.
Co-Authored-By: Brian Warner <[email protected]>
turadg
force-pushed
the
7502-reincarnate-feemint
branch
from
d639f0d
to
c537f6c
Compare
closes: #7502
Description
ZCF mint types were failing to restore during reincarnation. @warner traced this down to a faulty argument to
zcfMintBaggage.add(). That should have been caught by TypeScript but wasn't due to #7507There was an additional bug in awaiting a remote call (zoe vat from zcf vat) during reincarnation. This precludes the call during reincarnation by keeping the result in baggage. This changed allowed for less async handling so this also cleans up some related paths.
Security Considerations
issuer record in baggage
Scaling Considerations
--
Documentation Considerations
--
Testing Considerations
Existing tests pass. No unit tests for regression because it will be stressed by contract upgrade tests such as for #7466 in which this problem was discovered.