test(cosmic-swingset): pay 10 BLD for account with 0.25 IST to start

[dckc]

refs: #6067

Description

Integration test, including golang cosmos-sdk layers, to follow #6157

  ✔ PSM bootstrap and provision (1m 39.3s)
    ℹ make scenario2-run-psm-halt
    ℹ Fund pool with USDC
    ℹ make ACCT_ADDR=agoric1megzytg65cyrgzs6fvzxgrcqvwwl7ugpt62346 SOLO_COINS=1234000000ibc/usdc1234 fund-acct
    ℹ Fund user account with some BLD
    ℹ make ACCT_ADDR=agoric1ctncw9sguzftds55tss72n5htdla8e3s2wqrwt SOLO_COINS=123000000ubld fund-acct
    ℹ Provision an account
    ℹ make ACCT_ADDR=agoric1ctncw9sguzftds55tss72n5htdla8e3s2wqrwt provision-acct
    ℹ verify 10BLD spent, 0.25 IST received

Security Considerations

?

Documentation Considerations

?

Testing Considerations

increases ava timeout to 25 min in packages/cosmic-swingset

• TODO: wait until the 0.25 IST deposit should clear
• figure out why it fails with EOF in ci

[turadg]

Great to have this test automated!

[dckc]

Great to have this test automated!

yes... well, it would be... if it worked in ci like it does for me locally. But it doesn't, and the diagnostic doesn't give me much to go on:

Error: post failed: Post "http://localhost:26657": EOF
Usage:
  agd tx swingset provision-one [nickname] [address] [power-flags] [flags]

I'll need help in order to advance this.

[michaelfig]

Hi @dckc, do you want to try resurrecting this one and updating it to master? Or has its time passed?

[dckc]

Let's see... yes, this is a documented feature, so it should have an automated test.

As I mentioned, I'll need help. Do you have any clues for me, @michaelfig ?

Here's hoping for time to look at it together.

[dckc]

after spending about 2.5 days on this last week without getting a reliable result, how about we keep the resulting test code, which I mostly like, but use test.skip to keep the new test out of ci? I'm pushing changes to that effect.

[turadg]

how about we keep the resulting test code, which I mostly like, but use test.skip to keep the new test out of ci? I'm pushing changes to that effect.

Works for me, but before merge I'd like to get clear on:

1. Whether this is worth making work in CI eventually. If so, let's have an issue an reference it here.
2. How this will be tested outside of CI. What sorts of changes should trigger someone to uncomment .skip and disable the other one so it's testing?

It seems like if the problem is the coordination between the two tests, we could solve that by separating them and having two chain starts. That's prohibitive for CI but maybe not for the slow tests for merging to master. Alternately we could have a script like "run this locally when you change something in the econ boot". Eventually we could have CI be smart enough to run that slow one but only when necessary.

[turadg]

Seems tackling someday. If you agree, this would be a good place to reference an issue for it

[dckc]

ok: #6766

[dckc]

It seems like if the problem is the coordination between the two tests, we could solve that by separating them and having two chain starts.

That prompted me to reconsider how this is factored. I factored out scenario2.js and moved test-provision-smartwallet.js to its own test file... still with test.skip. This suggests 2 chain starts, but we can move the test back once we figure out the quirks.

It makes the change to test-make.js a bit less invasive... you don't have to wade thru as much provisionpool stuff to read that test.

That's prohibitive for CI but maybe not for the slow tests for merging to master.

I'm not familiar with how the slow tests work, so I lean toward leaving that aside.

2. How this will be tested outside of CI. What sorts of changes should trigger someone to uncomment .skip and disable the other one so it's testing?

Not sure... at least you and I know about it... I think of #6766 representing the work to make sure it's part of the development process.

[turadg]

Refactoring looks good. Thanks for the documentation and the new issue