[eth] Add Pyth Accumulator

[ali-behjati]

This PR adds the support WormholeMerkle accumulator message to the ethereum contract while still supporting the old message format. The code is not optimized yet and with more optimizations we can achieve a better gas usage. Currently based on the gas benchmark below it has a 18% improvement with a single price feed. Although the cost of updating 5 feeds in the same batch is higher than the current approach but in reality the chances that all 5 feeds be in the same batch is very low.

In the future when the migration to the Merkle Accumulator is over we can break the backward compatibility and remove the old code.

This is the result of the benchmark:

testBenchmarkGetEmaPrice() (gas: 25525)
testBenchmarkGetPrice() (gas: 23386)
testBenchmarkGetUpdateFee() (gas: 133310)
testBenchmarkParsePriceFeedUpdatesForOnePriceFeed() (gas: 310765)
testBenchmarkParsePriceFeedUpdatesForOnePriceFeedNotWithinRange() (gas: 311840)
testBenchmarkParsePriceFeedUpdatesForTwoPriceFeed() (gas: 317491)
testBenchmarkUpdatePriceFeedsIfNecessaryWhBatchFresh() (gas: 378912)
testBenchmarkUpdatePriceFeedsIfNecessaryWhBatchNotFresh() (gas: 175765)
testBenchmarkUpdatePriceFeedsWhBatchFresh() (gas: 357741)
testBenchmarkUpdatePriceFeedsWhBatchNotFresh() (gas: 316172)
testBenchmarkUpdatePriceFeedsWhMerkle1FeedFresh() (gas: 293517)
testBenchmarkUpdatePriceFeedsWhMerkle1FeedNotFresh() (gas: 283079)
testBenchmarkUpdatePriceFeedsWhMerkle2FeedsFresh() (gas: 333020)
testBenchmarkUpdatePriceFeedsWhMerkle2FeedsNotFresh() (gas: 312169)
testBenchmarkUpdatePriceFeedsWhMerkle3FeedsFresh() (gas: 359739)
testBenchmarkUpdatePriceFeedsWhMerkle3FeedsNotFresh() (gas: 339052)
testBenchmarkUpdatePriceFeedsWhMerkle4FeedNotFresh() (gas: 366044)
testBenchmarkUpdatePriceFeedsWhMerkle4FeedsFresh() (gas: 397336)
testBenchmarkUpdatePriceFeedsWhMerkle5FeedsFresh() (gas: 434540)
testBenchmarkUpdatePriceFeedsWhMerkle5FeedsNotFresh() (gas: 393001)

Note: This change set adds an overhead of 1k gas to the batch attestations.
Note 2: Gas benchmark here does not include the initial input size cost and that will be lower too. So I can say the gas saving would be slightly more than 20%.

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

2 Ignored Deployments

Name	Status	Preview	Comments	Updated (UTC)
example-oracle-amm	⬜️ Ignored (Inspect)			May 3, 2023 1:52pm
xc-admin-frontend	⬜️ Ignored (Inspect)			May 3, 2023 1:52pm

[jayantk]

message looks good except this one minor thing

[jayantk]

in general, this looks great. Nice job on very thorough testing. I left a couple minor comments

[jayantk]

will have to update this for the empty node prefix as well

[jayantk]

hmm yeah maybe we should charge on a per-feed basis, though doing that gas efficiently may be tricky. we may be able to do something a little too clever by using the number of bytes in an update as a proxy....

[ali-behjati]

I think we should do it per price and i like the idea of using number of bytes. What's wrong if i say len(update) * baseFee? 🤔

[ali-behjati]

Merging this PR. There are two outstanding TODOs for the future: (1) update fee (2) whether to emit an accumulator update or not.