Updates subscription tests for faster execution #211
Conversation
|
Awesome, looks much nicer now! I understand we don't need the timeouts anymore because the CI is executed in our machine? |
| function web3Subscribe(type: "logs", params: {}): Subscription<Log>; | ||
| function web3Subscribe(type: "newBlockHeaders" | "pendingTransactions" | "logs", params?: any) { | ||
| return (context.web3.eth as any).subscribe(...arguments); | ||
| } |
There was a problem hiding this comment.
What is the purpose of this code above? Is it to make sure the result of this function is typed for every type of event?
There was a problem hiding this comment.
You are right, I mostly meant the Typings is broken.
I defined this web3Subscribe to be able to return the modified Subscription class which allow the use of "once" (which is not declared in the default web3 subscription)
There was a problem hiding this comment.
Looks clean 👍
Probably when we use --dev which will make boot a lot faster
I'm not really expecting the node start time to get faster with #204
No, it is more related how the callback and promises were used. I tried my branch on the github servers (see https://github.com/PureStake/moonbeam/actions/runs/518123822) and it worked fine too. What I suspect is the fact that you provided a callback to the test, which in return prevents the use of async promises. But I didn't test the previous code on github servers so I'm not sure. |
The idea is that using the --dev instead of the plain specs, the load time will be a lot faster (already observed when doing that manually) |
What does it do?
It improves the subscription tests to avoid waiting 10s per test.
What important points reviewers should know?
web3 SDK is broken and types are not well implemented :/
Is there something left for follow-up PRs?
Some of the subscription tests are not independent (which is also true for many other tests to increase speed). This should be fixed at some point. (Probably when we use --dev which will make boot a lot faster)
What alternative implementations were considered?
Are there relevant PRs or issues in other repositories (Substrate, Polkadot, Frontier, Cumulus)?
What value does it bring to the blockchain users?
Checklist