Repair u19 tests in a3p-integration #10947

Chris-Hibbert · 2025-02-05T17:31:17Z

Description

In clean-up after upgrade-18, the upgrade 19 tests were left in an incomplete state. This ensures they are running again.

Security Considerations

Not a security issue.

Scaling Considerations

No implications.

Documentation Considerations

N/A

Testing and Upgrade Considerations

Our practice is to have a3p-integration represent upcoming upgrades to test them in the context of previous upgrades on chain. There were issues getting the U19 tests in proper condition before U18, and so revising them once U18 had been added to the chain was a little wonky. After this PR, The tests demonstrate that U19 will succeed, and the acceptance tests will verify the state of the chain after the upgrade.

cloudflare-workers-and-pages · 2025-02-05T21:00:44Z

Deploying agoric-sdk with Cloudflare Pages

Latest commit:	`f302496`
Status:	✅ Deploy successful!
Preview URL:	https://01a3dfe1.agoric-sdk.pages.dev
Branch Preview URL:	https://10901-repair-u19.agoric-sdk.pages.dev

View logs

Chris-Hibbert · 2025-02-05T23:18:12Z

@turadg, j@dckc, I've added you both as reviewers but it should only take one of you. Do you want to decide which is more available?

a3p-integration/proposals/n:upgrade-next/assetReserve.test.js

turadg · 2025-02-06T02:00:01Z

a3p-integration/proposals/n:upgrade-next/initial.test.js

@@ -10,7 +10,7 @@ const vats = {
  orchestration: { incarnation: 1 },
  transfer: { incarnation: 2 },
  walletFactory: { incarnation: 6 },
-  zoe: { incarnation: 4 },
+  zoe: { incarnation: 3 },


how did this number go down?

I don't know for sure how the test passed with this number. My guess would be that the code last ran in an environment where a Zoe upgrade was included with U18 or the proto U19.

The lower number is correct following U18, and in a U19 that doesn't upgrade Zoe.

I think because upgrade.go somehow kept some u18 proposals around when it was rebased on top of u18.

turadg · 2025-02-06T02:00:26Z

a3p-integration/proposals/n:upgrade-next/mint-payment/send-script.tjs

@@ -24,8 +24,6 @@ const sendBankAsset = async powers => {
  const valueStr = '{{VALUE}}';
  const value = BigInt(valueStr)

-  console.log(`Start sendBankAsset for ${label}`);


seems useful but you'd know better since you've been debugging

turadg · 2025-02-06T02:00:56Z

a3p-integration/proposals/n:upgrade-next/mintHolder.test.js

 } from './test-lib/mintHolder-helpers.js';
 import { networkConfig } from './test-lib/index.js';

-test('mintHolder contract is upgraded', async t => {
+test('verify mintHolder contract upgrade', async t => {


nit: "test" is already a verb

Hmm. I read the string argument as the name of the test, rather than a noun phrase describing what is to be tested. In this case the test used to perform the upgrade (because the proposal was a coreEval), and now only verifies it (because it's now a software upgrade).

a3p-integration/proposals/n:upgrade-next/mintHolder.test.js

turadg · 2025-02-06T02:09:57Z

a3p-integration/proposals/n:upgrade-next/package.json

      "testing/add-USD-OLIVES.js agoricNamesCoreEvals/addUsdOlives",
-      "testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo",
-      "vats/upgrade-mintHolder.js upgrade-mintHolder A3P_INTEGRATION"
+      "testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo"


all these proposals for testing should be under test/ so they don't invalidate the "use" image. see,

copy 'test' files only in TEST stage agoric-3-proposals#213

Making that change affects .gitignore and .tsconfig but if you put them under a path like test/submissions then you can use that pathname for those files and not have to manage a list of submissions.

We'll make this easier eventually,

remove sdk-generate build step #10537

Actually they should be in host, and their core eval output should be in test

Something like

Suggested change

"testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo"

"host/testing/publish-test-info.js test/agoricNamesCoreEvals/publishTestInfo"

FYI @Chris-Hibbert , the host thing hasn't landed in a3p. Up to you whether to wait for it to be merged and released to NPM so you can include that synthetic-chain lib version in this PR.

feat: proposal pre/post test steps agoric-3-proposals#210

I like your test/generated approach. It still copies to the image unnecessarily, but it's only the test image so it doesn't invalid any extra. (If the proposal changes its build would anyway). I think it's better to keep test stuff together than to spread it into another path.

Yeah you're right, a change in the proposal building would likely invalidate everything anyway, but collocating test files together is cleaner.

I got confused about the host stuff however, I forgot these first argument is relative to the SDK packages.

turadg · 2025-02-06T02:13:43Z

golang/cosmos/app/upgrade.go

-// 	if variant == "" {
-// 		return nil, nil
-// 	}
+func restartFeeDistributorCoreProposal(targetUpgrade string) (vm.CoreProposalStep, error) {


I'm not reviewing any of the Golang in this PR

packages/builders/scripts/inter-protocol/replace-feeDistributor-combo.js

turadg · 2025-02-06T02:14:57Z

packages/builders/scripts/inter-protocol/replace-feeDistributor-combo.js

+  });
+};
+
+const Usage = `agoric run replace-feedDistributor-combo.js ${[...knownVariants, '<json-config>'].join(' | ')}`;


surprising casing.

it matches replace-feeDistributor.js.

I don't have an objection to this form. If you want something different, please say what would be better.

turadg · 2025-02-06T02:19:13Z

packages/builders/scripts/inter-protocol/replace-feeDistributor-combo.js

+/** @type {import('@agoric/deploy-script-support/src/externalTypes.js').DeployScriptFunction} */
+export default async (homeP, endowments) => {
+  const { scriptArgs } = endowments;
+  const variantOrConfig = scriptArgs?.[0];


I'm not familiar with this convention. Please factor it out into a reusable function with documentation.

mhofman

It's a little unfortunate we have more need for variants. But since that's already the case for the mintHolder I won't block on that.

mhofman · 2025-02-07T00:02:11Z

a3p-integration/proposals/n:upgrade-next/package.json

      "testing/add-USD-OLIVES.js agoricNamesCoreEvals/addUsdOlives",
-      "testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo",
-      "vats/upgrade-mintHolder.js upgrade-mintHolder A3P_INTEGRATION"
+      "testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo"


Actually they should be in host, and their core eval output should be in test

Something like

Suggested change

"testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo"

"host/testing/publish-test-info.js test/agoricNamesCoreEvals/publishTestInfo"

mhofman · 2025-02-07T00:04:38Z

a3p-integration/proposals/n:upgrade-next/psm.test.js

@@ -29,38 +27,19 @@ test.before(async t => {
  };
 });

-test.serial('simulate trade of IST and USDC', async t => {
+test.serial('check stats pre-swap', async t => {


I won't block on this, but we've been abusing test.serial. Its only guarantee is "not serial" and we've been relying on it being sequential.

You mean "not concurrent". I honestly didn't realize it didn't guarantee sequential.

mhofman · 2025-02-07T00:09:32Z

golang/cosmos/app/upgrade.go

+		builderScript,
+		"defaultProposalBuilder",
+		map[string]any{
+			"variant": variant,


Really unfortunate we still have the need for variants in this upgrade. We need to find a way out of this pickle, but I'm not gonna try to figure out why this upgrade needs special casing per network now.

In the mean time, we can hopefully prioritize #10837

Edit: filed #10956

I've read #10837 and don't understand the path forward.

This distinction is because we want to be able to validate the feeDistributor, and waiting a full hour-long cycle for it seems infeasible. We have other timing-related variations for auctions. In order to get rid of variant configurations for test, we need another solution.

I think my core question was why these config have to be provided as core proposal argument, and cannot be stored in the config of the contract we're upgrading. That's what #10956 asks. We can punt on it for now.

mhofman · 2025-02-07T00:10:03Z

golang/cosmos/app/upgrade.go

+            	// because of #10794, we need to do at least a null upgrade of
+            	// the walletFactory on every software upgrade


Looks like the comments needs formatting (weird indentation)

mhofman · 2025-02-07T00:11:17Z

a3p-integration/proposals/n:upgrade-next/initial.test.js

@@ -10,7 +10,7 @@ const vats = {
  orchestration: { incarnation: 1 },
  transfer: { incarnation: 2 },
  walletFactory: { incarnation: 6 },
-  zoe: { incarnation: 4 },
+  zoe: { incarnation: 3 },


I think because upgrade.go somehow kept some u18 proposals around when it was rebased on top of u18.

packages/builders/scripts/inter-protocol/replace-feeDistributor-combo.js

mhofman · 2025-02-07T00:15:30Z

packages/builders/scripts/inter-protocol/replace-feeDistributor-combo.js

+import { getManifestForReplaceFeeDistributor } from '@agoric/inter-protocol/src/proposals/replace-fee-distributor.js';
+import { SECONDS_PER_HOUR } from '@agoric/inter-protocol/src/proposals/econ-behaviors.js';
+
+const configurations = {


So I'll bite. Why do these config need to be hard coded in the builder? Can't they be restored from whatever params are currently on the respective chain, removing the need for variants?

The feeDistributor in a3p-integration currently has the same parameter settings as on mainNet. We want to run tests that verify that fees are distributed. Without a change, the test would have to wait an hour. We make similar changes with changes to timing for tests for other vats, like auctions.

And we can't update these parameters outside of the core proposal, through a core eval or other governance vote, either before the upgrade, or after in the test?

generated proposals into .../generated

turadg · 2025-02-07T23:25:56Z

a3p-integration/proposals/n:upgrade-next/test/test-lib/mintHolder-helpers.js

@@ -62,6 +62,12 @@ export const mintPayment = async (t, address, assetList, value) => {

  for (const asset of assetList) {
    const { label, denom } = asset;
+
+    // XXX I don't know what asset shows up like this, but let's ignore it.


please indicate why it's okay to ignore

turadg · 2025-02-07T23:30:17Z

a3p-integration/proposals/n:upgrade-next/package.json

      "testing/add-USD-OLIVES.js agoricNamesCoreEvals/addUsdOlives",
-      "testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo",
-      "vats/upgrade-mintHolder.js upgrade-mintHolder A3P_INTEGRATION"
+      "testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo"


I like your test/generated approach. It still copies to the image unnecessarily, but it's only the test image so it doesn't invalid any extra. (If the proposal changes its build would anyway). I think it's better to keep test stuff together than to spread it into another path.

Chris-Hibbert added deployment Chain deployment mechanism (e.g. testnet) test Testnet contract-upgrade labels Feb 5, 2025

Chris-Hibbert self-assigned this Feb 5, 2025

Chris-Hibbert requested a review from a team as a code owner February 5, 2025 17:31

Chris-Hibbert requested a review from AgoricTriage February 5, 2025 17:31

Chris-Hibbert marked this pull request as draft February 5, 2025 17:31

Chris-Hibbert added 5 commits February 5, 2025 12:59

test: migrate u19 tests to upgrade-next; update upgrade.go

750a7f2

test: repair feeDistributor test: allow config from upgrade.go

5af83d0

test: repair mintHolder test: add sendScript & exclude fastUSDC

b068711

test: drop redundant bundles in assetReserve test

af22f09

test: update PSM test

f7dfbac

Chris-Hibbert force-pushed the 10901-repair-U19 branch from 6b1d701 to f7dfbac Compare February 5, 2025 21:00

Chris-Hibbert added the force:integration Force integration tests to run on PR label Feb 5, 2025

Chris-Hibbert marked this pull request as ready for review February 5, 2025 21:01

chore: drop an unused import

420220f

Chris-Hibbert requested review from turadg and dckc February 5, 2025 23:16

turadg reviewed Feb 6, 2025

View reviewed changes

mhofman reviewed Feb 7, 2025

View reviewed changes

turadg mentioned this pull request Feb 7, 2025

feat: proposal pre/post test steps Agoric/agoric-3-proposals#210

Merged

Chris-Hibbert added 2 commits February 7, 2025 15:07

test: move a3p-integration test files under .../test/

5576d7e

generated proposals into .../generated

refactor: clean-ups in proposals

d628279

turadg approved these changes Feb 7, 2025

View reviewed changes

Chris-Hibbert added 3 commits February 7, 2025 16:01

refactor: extract common method for parsing scriptArgs

17f2c0d

chore: add configurations for Devnet and Emerynet for feeDistributor

730d4f5

chore: gofmt upgrade.go

f302496

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repair u19 tests in a3p-integration #10947

Repair u19 tests in a3p-integration #10947

Chris-Hibbert commented Feb 5, 2025

cloudflare-workers-and-pages bot commented Feb 5, 2025 •

edited

Loading

Chris-Hibbert commented Feb 5, 2025

turadg Feb 6, 2025

Chris-Hibbert Feb 6, 2025

mhofman Feb 7, 2025

turadg Feb 6, 2025

turadg Feb 6, 2025

Chris-Hibbert Feb 6, 2025

turadg Feb 6, 2025

mhofman Feb 7, 2025

turadg Feb 7, 2025

turadg Feb 7, 2025

mhofman Feb 8, 2025

turadg Feb 6, 2025

turadg Feb 6, 2025

Chris-Hibbert Feb 8, 2025

turadg Feb 6, 2025

Chris-Hibbert Feb 8, 2025

mhofman left a comment

mhofman Feb 7, 2025

mhofman Feb 7, 2025

mhofman Feb 7, 2025 •

edited

Loading

Chris-Hibbert Feb 8, 2025

mhofman Feb 8, 2025

mhofman Feb 7, 2025

Chris-Hibbert Feb 8, 2025

mhofman Feb 7, 2025

mhofman Feb 7, 2025

Chris-Hibbert Feb 8, 2025

mhofman Feb 8, 2025

turadg Feb 7, 2025

turadg Feb 7, 2025

	"testing/publish-test-info.js agoricNamesCoreEvals/publishTestInfo"
	"host/testing/publish-test-info.js test/agoricNamesCoreEvals/publishTestInfo"

		// because of #10794, we need to do at least a null upgrade of
		// the walletFactory on every software upgrade

Repair u19 tests in a3p-integration #10947

Are you sure you want to change the base?

Repair u19 tests in a3p-integration #10947

Conversation

Chris-Hibbert commented Feb 5, 2025

Description

Security Considerations

Scaling Considerations

Documentation Considerations

Testing and Upgrade Considerations

cloudflare-workers-and-pages bot commented Feb 5, 2025 • edited Loading

Deploying agoric-sdk with Cloudflare Pages

Chris-Hibbert commented Feb 5, 2025

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

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

mhofman left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mhofman Feb 7, 2025 • edited Loading

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

Choose a reason for hiding this comment

cloudflare-workers-and-pages bot commented Feb 5, 2025 •

edited

Loading

mhofman Feb 7, 2025 •

edited

Loading