feat: durable agoricNames, namesByAddress #7605

dckc · 2023-05-04T06:54:48Z

Description

durable makeNameHubKit with NameAdmin.provideChild()
move namesByAddress kit to vat-provisioning.js
move agoricNames kit to its own vat, vat-agoricNames.js
provideInertBrand() in vat-agoricNames to move oracleBrands out of bootstrap
eliminate a couple boot strap Far attenuators that were not worthwhile

Security Considerations

Reduce responsibilities of the bootstrap vat; after moving these two name services, after bootstrap, the bootstrap vat should only be used for core eval.

Scaling Considerations

namesByAddress is now virtualized and durable and out of the bootstrap vat

Documentation Considerations

none?

Testing Considerations

normal-to-light unit testing

turadg · 2023-05-06T13:11:49Z

packages/vats/src/core/lib-boot.js

@@ -163,7 +163,7 @@ export const makeBootstrap = (
          vatData.set(name, { root });
        }
      }
-      void rawBootstrap(vats, devices).catch(e => {
+      return rawBootstrap(vats, devices).catch(e => {


turadg · 2023-05-06T13:13:05Z

packages/inter-protocol/src/vaultFactory/types.js

@@ -108,7 +108,7 @@
 * @typedef {object} LiquidationStrategy
 * @property {() => KeywordKeywordRecord} keywordMapping
 * @property {(collateral: Amount, run: Amount) => Proposal} makeProposal
- * @property {(runDebt: Amount) => Promise<Invitation>} makeInvitation
+ * @property {(debt: Amount) => Promise<Invitation>} makeInvitation


I pushed this stray commit by accident. It doesn't hurt but if you rebase please excise

packages/inter-protocol/test/smartWallet/boot-psm.js

packages/internal/test/test-callback.js

packages/vats/src/core/chain-behaviors.js

turadg · 2023-05-06T13:55:29Z

packages/vats/src/nameHub.js

-    keys: () => {
-      return harden([...keyToRecord.keys()]);
-    },
+export const prepareNameHubKit = (zone = heapZone) => {


a prepare without a zone argument doesn't make sense to me. you only ever make a prepare if you're eventually making it durable. I think we should require the zone.

Moreover to avoid dangling baggages I think the non-durable zones should warn when kinds are defined into them. It can only be an interim state during development. À propos,

Audit makeScalarBigMapStore usages for pseudo baggage #7578

a prepare without a zone argument doesn't make sense to me.

agreed. changing.

I think the non-durable zones should warn when kinds are defined into them.

good idea; not for this PR, OK?

turadg · 2023-05-06T13:57:18Z

packages/vats/src/nameHub.js

+  const init1 = () => ({
+    /** @type {LegacyMap<string, PromiseKit<unknown>>} */
+    // Legacy because a promiseKit is not a passable
+    keyToPK: makeLegacyMap('nameKey'),


Then use JS Map. LegacyMap is not to be used for any new code. It's scheduled for deletion.

turadg · 2023-05-06T14:00:11Z

packages/vats/src/nameHub.js

+          /** @type {any} */
+          const firstValue = keyToValue.has(first)
+            ? keyToValue.get(first)
+            : keyToPK.get(first).promise; // or throw


JS Map doesn't throw on missing values. To get that back a clean option is our NonNullish checked cast:

Suggested change

: keyToPK.get(first).promise; // or throw

: NonNullish(keyToPK.get(first)).promise;

turadg · 2023-05-06T14:07:22Z

packages/vats/src/vat-agoricNames.js

+
+/** @param {import('@agoric/zone').Zone} zone */
+const makeBrandStore = zone => {
+  const brandStore = zone.mapStore('Brand');


Took me a minute to see this wasn't making singletons. I'm still getting used to zones.

One thing that would help readability is moving const brandStore below the kind definition. Kinds should come first. Anything that the kind def depends on should be obvious so it's clear what has to happen in the first crank of restart.

right... refactored to just prepareNatBrand, without a closure for the state.

turadg · 2023-05-06T14:07:58Z

packages/vats/src/vat-agoricNames.js

+  );
+
+  return {
+    provide: (keyword, displayInfo) =>


The plain provide is for when what is provided is specified by the make argument. In this case it's always making a brand (and that's not implied by the signature).

Suggested change

provide: (keyword, displayInfo) =>

provideBrand: (keyword, displayInfo) =>

turadg

(after squashing the fixups and reverts)

turadg · 2023-05-06T23:48:25Z

packages/vats/test/bootstrapTests/test-vaults-upgrade.js

@@ -67,7 +67,7 @@ const makeDefaultTestContext = async t => {
 test.before(async t => {
  t.context = await makeDefaultTestContext(t);
 });
-test.after.always(t => t.context.shutdown());
+test.after.always(t => t.context.shutdown?.());


this seems like a bad idea to me. Every bootstrap test needs to shutdown the host storage.

- provideChild is async - update is _not_ async

home.myAddressNameAdmin is a NameAdmin with an extra getMyAddress() method. Here we use exo "attenuators" to - add getMyAddress() to a NameAdmin, making a MyAddressNameAdmin - override provideChild() in NamesByAddressAdmin, providing a kit whose nameAdmin is a MyAddressNameAdmin This obsoletes makeMyAddressNameAdmin and simplifies a previously delicate and fragile coordination between the smart wallet and the ag-solo wallet.

The datalock from pismo seems to be gone. This should facilitate diagnostics.

Co-authored-by: Turadg Aleahmad <turadg@agoric.com>

dckc force-pushed the dc-durable-namehub branch from fb91ba4 to 4cc9fbb Compare May 4, 2023 18:07

dckc force-pushed the dc-space-reserve-any branch from 2f06b7e to 4e9bad6 Compare May 4, 2023 18:58

dckc marked this pull request as draft May 4, 2023 19:08

dckc force-pushed the dc-durable-namehub branch from 4cc9fbb to c7d9760 Compare May 4, 2023 21:58

dckc force-pushed the dc-space-reserve-any branch 2 times, most recently from b24025b to 717a002 Compare May 5, 2023 04:47

dckc force-pushed the dc-durable-namehub branch 7 times, most recently from 58f003e to 43ec75d Compare May 5, 2023 17:48

dckc changed the title ~~WIP: toward durable namehub~~ WIP: 4548 durable bootstrap May 5, 2023

turadg force-pushed the dc-space-reserve-any branch from 206d58c to 44fdc31 Compare May 5, 2023 18:15

dckc force-pushed the dc-durable-namehub branch from 43ec75d to 1d0cb37 Compare May 5, 2023 19:04

Base automatically changed from dc-space-reserve-any to master May 5, 2023 21:11

dckc force-pushed the dc-durable-namehub branch 4 times, most recently from 18ed498 to 4db2eb2 Compare May 6, 2023 05:51

turadg reviewed May 6, 2023

View reviewed changes

dckc force-pushed the dc-durable-namehub branch 3 times, most recently from 66c4cd7 to 9cb0057 Compare May 6, 2023 20:29

dckc changed the title ~~WIP: 4548 durable bootstrap~~ feat: durable agoricNames, namesByAddress May 6, 2023

dckc force-pushed the dc-durable-namehub branch from b2d06fd to e21dc09 Compare May 6, 2023 20:53

dckc marked this pull request as ready for review May 6, 2023 21:00

dckc requested a review from turadg May 6, 2023 21:01

turadg approved these changes May 6, 2023

View reviewed changes

michaelfig and others added 24 commits May 7, 2023 04:19

fix(internal): more callback typing

e78e14d

feat: nameHub: mixinMyAddress

6c305c4

docs: note on 2 calls to createVatAdminService

7cc2904

test(inter-protocol): type fixes (from ???)

8ffef18

test(callback): prune ts-expect-error

3a54887

feat: durable namehub with .has(), .entries(false)

11e61b5

- provideChild is async - update is _not_ async

test(wallet-api): update makeFakeMyAddressNameAdmin mock

0456a97

fixup(nameHub): don't skip promises for onUpdate callback

75819f3

feat: durable agoricNames in its own vat

526396e

feat(zone): check baggage

62e8a75

feat(provisioning): durable vat for namesByAddress

5d32752

chore: show agoricNames vat in addBankAssets permit

e0f906a

chore: address cross-package lint:types strangeness

17146aa

docs: make SOLO_SLOGFILE greppable

7c3ce54

chore: provide agoricNames bundle to solo-config

2cb562b

fix: don't drop the rawBootstrap promise

90a742a

The datalock from pismo seems to be gone. This should facilitate diagnostics.

chore: postpone durable storage for makeVatSpace

8f2254b

fixup: clarify chainStorage diagnostic

b0303f4

Co-authored-by: Turadg Aleahmad <turadg@agoric.com>

chore: zone not optional in prepare

661681a

chore: use Map rather than makeLegacyMap

8c239d8

refactor: prepareNatBrand in provideInertBrand

6385ead

test: regen boot viz snapshots

b733f8e

chore: vatSpace: Only remotables can be keys of scalar WeakMapStores

a6f85f7

dckc force-pushed the dc-durable-namehub branch from 31afd15 to a6f85f7 Compare May 7, 2023 04:19

dckc added this pull request to the merge queue May 7, 2023

Merged via the queue into master with commit 0ce9545 May 7, 2023

dckc deleted the dc-durable-namehub branch May 7, 2023 12:39

mhofman mentioned this pull request May 8, 2023

No fake baggage #7636

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: durable agoricNames, namesByAddress #7605

feat: durable agoricNames, namesByAddress #7605

dckc commented May 4, 2023 •

edited

Loading

turadg May 6, 2023

turadg May 6, 2023

turadg May 6, 2023

dckc May 6, 2023

dckc May 6, 2023

turadg May 6, 2023

turadg May 6, 2023

turadg May 6, 2023

dckc May 6, 2023

turadg May 6, 2023

turadg left a comment

turadg May 6, 2023

dckc May 7, 2023

	: keyToPK.get(first).promise; // or throw
	: NonNullish(keyToPK.get(first)).promise;

	provide: (keyword, displayInfo) =>
	provideBrand: (keyword, displayInfo) =>

feat: durable agoricNames, namesByAddress #7605

feat: durable agoricNames, namesByAddress #7605

Conversation

dckc commented May 4, 2023 • edited Loading

Description

Security Considerations

Scaling Considerations

Documentation Considerations

Testing Considerations

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

turadg left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

dckc commented May 4, 2023 •

edited

Loading