chore(a3p-integration): Improve log output

[gibson042]

Description

Extracted from #10165 and best reviewed by commit.

• Include relevant variable details (e.g., block height).
• Omit output that doesn't convey new information "done", "reading", etc.).
• Provide a common logging prefix for subtasks, inlining functions where relevant.
• Remove endo/init/legacy.js from z:acceptance ava config to eliminate noisy Object <[Object: null prototype] {}> output.
• Introduce a logRecord helper to concisely log possibly-remotable-bearing records and/or record entries.

Security Considerations

n/a

Scaling Considerations

n/a

Documentation Considerations

n/a

Testing Considerations

n/a

Upgrade Considerations

n/a

[cloudflare-workers-and-pages]

Deploying agoric-sdk with    Cloudflare Pages

Latest commit:	5b2e20c
Status:	✅  Deploy successful!
Preview URL:	https://458fdf60.agoric-sdk.pages.dev
Branch Preview URL:	https://gibson-2024-11-a3p-integrati.agoric-sdk.pages.dev

View logs

[dckc]

I'm afraid I don't know enough to approve.

[dckc]

I wonder why the denominator is not 100n. But I suppose that pre-dates this PR?

[gibson042]

Yep, I just moved it. 🤷

[dckc]

I'm not confident that I can follow this logic. And I'm not quite sure what it's supposed to do.

I'm looking for examples in ci logs, but I see:

Integration tests / test-docker-build (pull_request) Skipped

... even though the force:integration label is present.

Some examples in the docstring might be nice. Or unit tests.

[gibson042]

Updated the docstring. Actual output is visible in a prior CI run: https://github.com/Agoric/agoric-sdk/actions/runs/11881729913/job/33106407896#step:10:4506

  ✔ swap into IST using agd with default gas (9s)
    ℹ METRICS  {
        anchorPoolBalance: {
          brand: Object @Alleged: BoardRemoteUSDC_axl brand {},
          value: 30010011n,
        },
        feePoolBalance: {
          brand: Object @Alleged: BoardRemoteIST brand {},
          value: 0n,
        },
        mintedPoolBalance: {
          brand: Object @Alleged: BoardRemoteIST brand {},
          value: 30010011n,
        },
        totalAnchorProvided: {
          brand: Object @Alleged: BoardRemoteUSDC_axl brand {},
          value: 0n,
        },
        totalMintedProvided: {
          brand: Object @Alleged: BoardRemoteIST brand {},
          value: 30010011n,
        },
      }
    ℹ BALANCES { [denom]: amount, ... } {
        'ibc/295548A78785A1007F232DE286149A6FF512F180AF5657780FC89C009E2C348F': '300000000',
        ubld: '20000000',
        uist: '250000',
      }

[turadg]

This is a lot of new code in psm-lib that isn't specific to PSM. I'm not opposed to landing it but please don't spend effort improving these deprecated utils. It would be better spent on @agoric/client-utils and getting psm-lib to use that.

incidentally, consider marking @deprecated all the functions in psm-lib that aren't about PSM per se.

[dckc]

I should know more about why we were using endo legacy before approving this.

I'm also not sure about why use imports in every file rather than ava require. I sort of prefer it that way, but it hasn't been the norm here.

[gibson042]

The general reason seems to generally be JS assignment override mistake mitigation, albeit in a noisy way:

$ for import in @endo/init @endo/init/legacy.js; do
  echo
  node --input-type=module -e "
    import '$import';
    import { inspect } from 'node:util';
    console.log('$import');
    let MyObject = () => 'not implemented';
    try {
      MyObject = function MyObject(props = {}) {
        return new.target ? Object.assign(this, props) : new MyObject(props);
      };
      (MyObject.prototype = Object.create(Object.prototype)).constructor = MyObject;
    } catch (err) { console.log('ERROR', err.message); }
    const instance = MyObject({ foo: 'bar' });
    console.log(util.inspect({
      data: {
        object: {instance},
        map: new Map([['instance', instance]]),
      }
    }, { depth: 10 }));
  "
done

@endo/init
ERROR Cannot assign to read only property 'constructor' of object '[object Object]'
{
  data: {
    object: { instance: { foo: 'bar' } },
    map: Map(1) { 'instance' => { foo: 'bar' } }
  }
}

@endo/init/legacy.js
Object <[Object: null prototype] {}> {
  data: Object <[Object: null prototype] {}> {
    object: Object <[Object: null prototype] {}> {
      instance: MyObject { foo: 'bar' }
    },
    map: Map <Object <[Object: null prototype] {}>>(1) [Map] {
      'instance' => MyObject { foo: 'bar' }
    }
  }
}

But AFAICT, it is not helpful here.

[turadg]

IIRC some import of axios or protobufjs prompted it. I didn't realize it had any DX downsides. Given that it impairs the console messages I agree we should avoid it.

Howeverplease keep the endo/init config in the Ava require so it's consistent across tests. Also, shouldn't it be @endo/init/debug.js for better output?

[gibson042]

shouldn't it be @endo/init/debug.js for better output?

Good point; done.

[turadg]

Let's not put any more effort into improving these functions that can be replaced by @agoric/client-utils.

[turadg]

This is a lot of new code in psm-lib that isn't specific to PSM. I'm not opposed to landing it but please don't spend effort improving these deprecated utils. It would be better spent on @agoric/client-utils and getting psm-lib to use that.

incidentally, consider marking @deprecated all the functions in psm-lib that aren't about PSM per se.

[turadg]

It's in ERTP now. Why wait?

Suggested change

	// TODO @import {Ratio} from '@agoric/zoe'
	/** @typedef {any} Ratio */
	/** @import {Ratio} from '@agoric/ertp`;

(it might need src/types.js)

[gibson042]

error TS2305: Module '"@agoric/ertp"' has no exported member 'Ratio'.

error TS2305: Module '"@agoric/ertp/src/types.js"' has no exported member 'Ratio'.

[turadg]

Ah, the proposal must have an old version of ERTP. We'd have the one in agoric-sdk if this used the new 'yarn link' trick but it's not worth further churn on the PR.