Update to wasmvm 2.0.0-rc.1

[chipshort]

Very WIP, main goal is to get it to compile and pass current tests

TODOs:

• supported capabilities are now []string
• nested Result changes
• renaming
• StoreCode gas
• ValidateAddress
• Array[C] type replacements
• Adjust gas multiplier
• TransferMsg.Memo
• QueryRequest.Grpc
• MsgResponses (see [idea, unfinished] Add flattened msgResponses to SubMsgResponse #1800)
• IBCReceiveResponse.Acknowledgement == nil
• Payload
• Make redactError call conditional in func (d MessageDispatcher) DispatchSubmessages

[chipshort]

@webmaster128 Any idea why the CI docker-image libwasmvm check is failing?
When I locally run (cd cmd/wasmd && go run . query wasm libwasmvm-version), I get 2.0.0-rc.1. Is the docker image an old version?

[webmaster128]

@webmaster128 Any idea why the CI docker-image libwasmvm check is failing?
When I locally run (cd cmd/wasmd && go run . query wasm libwasmvm-version), I get 2.0.0-rc.1. Is the docker image an old version?

I did not check the error yet but you need to update those lines for sure: https://github.com/CosmWasm/wasmd/blob/main/Dockerfile#L18-L21

[webmaster128]

I think it would be valuable to do as many of the Probably in follow-up PRs items in here as well. This PR then serves as a go-to-place for every other user of wasmvm (which right now is primarily wasmd forks and the Wasm Light Client).

[pinosu]

Changes look good so far! To make the CI pass you need to update the Dockerfile both version and checksum for libwasmvm.

[chipshort]

@pinosu @webmaster128 Does one of you know the difference between the different reflect.wasm files and reflect_1_1.wasm used in tests? Because of the Stargate to Any rename, we either need to update them to the CW 2.0 version or we need to continue calling them with stargate json. Any opinions on that?

[webmaster128]

I think it would be pretty cool to test both, the old (stargate) and the new (any) message type to ensure wasmd supports both. I don't know about the 1_1 one. it was added here: https://github.com/CosmWasm/wasmd/pull/1055/files

[webmaster128]

Could you add the cosmwasm_2_0 capability to the README (on top of #1805)?

[webmaster128]

Good stuff! Not done reviwwing but here is a start

[chipshort]

The test-system job seems to fail sporadically. Is it just a timing thing?

[pinosu]

The test-system job seems to fail sporadically. Is it just a timing thing?

Yes, unfortunately since we added the system test for the upgrade, sporadically it fails for timeout

[webmaster128]

If you look at the logs and the implementation of AwaitBlockHeight we see that we wait for block number 21, current block is 9 and block time is 1s with a timeout of 14s. Finalizing 12 one-second blocks within 14 seconds is very optimistic given that the blockTime parameter is used for "--commit-timeout=" + s.blockTime.String(),. But commit timeout is not the block time but a lower bound for the block time (https://docs.tendermint.com/v0.34/tendermint-core/configuration.html). For Nois we use e.g.

			// Consensus settings
			config.Consensus.TimeoutPropose = 2000 * time.Millisecond
			config.Consensus.TimeoutProposeDelta = 500 * time.Millisecond
			config.Consensus.TimeoutPrecommitDelta = 500 * time.Millisecond
			config.Consensus.TimeoutPrevote = 1 * time.Second
			config.Consensus.TimeoutPrecommit = 1 * time.Second
			config.Consensus.TimeoutPrecommitDelta = 500 * time.Millisecond
			config.Consensus.TimeoutCommit = 1750 * time.Millisecond

to get block times of 2.11s in production or

# Custom settings for very fast blocks in CI
# We target 250ms block times with one validator.
sed -i"" \
  -e 's/^timeout_propose =.*$/timeout_propose = "100ms"/' \
  -e 's/^timeout_propose_delta =.*$/timeout_propose_delta = "100ms"/' \
  -e 's/^timeout_prevote =.*$/timeout_prevote = "100ms"/' \
  -e 's/^timeout_prevote_delta =.*$/timeout_prevote_delta = "100ms"/' \
  -e 's/^timeout_precommit =.*$/timeout_precommit = "100ms"/' \
  -e 's/^timeout_precommit_delta =.*$/timeout_precommit_delta = "100ms"/' \
  -e 's/^timeout_commit =.*$/timeout_commit = "230ms"/' \
  "config/config.toml"

to aim for 250ms block times in CI. So I think the --commit-timeout should be something like 90% of the desired block time.

[webmaster128]

Very very nice solution

[webmaster128]

Do we have an integration test somewhere that shows how a submessage error is returned to a contract unredacted? This is a major feature in CosmWasm 2.0 which we should highlight and test, even if creating such a test is a bunch of work.

[chipshort]

Do we have an integration test somewhere that shows how a submessage error is returned to a contract unredacted?

Good point, I added a test now that instantiates two reflect instances and makes one send a submessage that fails to the other one. It then checks the error message the first contract got.

[pinosu]

Great work! 💯 LGTM 👍