Remove POC Zoe metering

[erights]

A step towards #4225 . You can review that one against master, which is probably more coherent. Or you can review these individually.

[dckc]

Did I miss a memo? Why remove Zoe metering?
Is there an issue I can look at for context?

[erights]

Did I miss a memo? Why remove Zoe metering? Is there an issue I can look at for context?

No. Came up first in a discussion I had with @warner. I then discussed it in this morning's eng sync. It's on the agenda for tomorrow's (oops, today's) Zoe meeting, where we'll discuss it. Could you be there? (we'll record in any case)

Summary:

• We don't need it for mainnet 1,
• it is a significant source of complexity, and
• removing it was almost purely subtractive.
• The removal caused no compensating engineering down a throw-away dead end.

The Least Authority review was pre-metering. We're about to go into a purple teaming effort on Zoe, where I'm supposed to be blue, and I do not know enough to try to defend that code. And it is a distraction from what we should be worried about for mainnet 1.

For mainnet 1, as we discussed at the all-hands, our plans to prevent DOS and runaways is vat-level, not contract-level anyway. Let's get this in place first and get some experience with it.

Finally, we plan to replace Zoe 1 with Zoe 2 after mainnet 1. That seems like the right time to take this on for production. Because Zoe 2 starts so much simpler, we're more likely to be able to add these additional concerns with higher confidence.

[Chris-Hibbert]

This looks good.

We should talk about the order of merging between this and my rearranging of the run-protocol contracts.

[dckc]

• We don't need it for mainnet 1,

How about for devnet? That is: how do smart contract developers get a feel for how much their contracts will cost to run?

[erights]

• We don't need it for mainnet 1,

How about for devnet? That is: how do smart contract developers get a feel for how much their contracts will cost to run?

We discussed in the kernel meeting. There isn't enough value here to make it worth keeping, given that:

• We need to have kernel/vat level metering and fees by mainnet 1 anyway.
• The current zoe API with expiry is wrong anyway (though as Dean points out this could be removed while leaving the rest intact)
• The important devnet zoe API will be zoe 2.

So we decided to move ahead with this PR together with #4225.

[erights]

This looks good.

We should talk about the order of merging between this and my rearranging of the run-protocol contracts.

We talked and decided that if this can go in today, which seems likely, then it should go first.

[erights]

#4225 was built on this one and was merged to master, making this PR irrelevant. Closing.