integrations: Remove extra fetching of runtime public key

[aefhm]

Description

Address Node timeout issue encountered by developers.

Im using the alpha of @oasisprotocol/sapphire-viem-v2, and it looks like when using it, node process wont naturally terminate (w/o a call to exit() ). I think that its the fault of this timer

sapphire-paratime/integrations/viem-v2/src/index.ts

Lines 126 to 128 in 77a2bea

	setTimeout(async () => {
	await fetcher.fetch(provider);
	}, fetcher.timeoutMilliseconds);

Resolution

I don't think we need the second fetch.

[netlify]

✅ Deploy Preview for oasisprotocol-sapphire-paratime canceled.

Name	Link
🔨 Latest commit	0dae328
🔍 Latest deploy log	https://app.netlify.com/sites/oasisprotocol-sapphire-paratime/deploys/66d8f14619c6290008c6431c

[CedarMist]

So, this is a workaround for the fact that Viem's code for packing transactions isn't async, meaning when we encrypt a transaction we can't do an RPC request to retrieve the calldata public key.

We do the first fetch to retrieve the public key, and then the timer proactively fetches public keys.

Without the timer and periodic public key fetching the app will eventually start encrypting transactions with expired keys causing it to fail.

It would be good to find a workaround for this, as it's not the best. But just removing it will cause long-running dApps to break.

[CedarMist]

It would be good to setup a test case for using Viem from Node CLI to reproduce.

And use https://www.npmjs.com/package/wtfnode to see which handles are still active.

Could use a WeakRef to allow the callback function to be garbage collected, but that's not 100% predictable, see:

let callback = () => {
  console.log('Interval tick');
  // Your interval logic here
};

const weakCallback = new WeakRef(callback);

const intervalId = setInterval(() => {
  const cb = weakCallback.deref();
  if (cb) {
    cb();
  } else {
    console.log('Callback has been garbage collected');
    clearInterval(intervalId);
  }
}, 1000);

// At some point later
callback = null; // Allow the callback to be garbage collected

Or we could trap SIGINT via process.on to clear the interval... Would need to detect if it's Node or Web via checking if globalThis.process is undefined.

Or we could unref the interval ID, e.g.

https://albraga.gitbooks.io/mastering-node-js-book/content/unref-and-ref.html

The unref method allows the developer to assert the following instructions: when this timer is the only event source remaining for the event loop to process, go ahead and terminate the process.

setTimeout(() => { console.log("now stop"); }, 100); 

let intervalId = setInterval(() => { console.log("running") }, 1); 

intervalId.unref()

[CedarMist]

Or... we could refactor the Viem code to not do any transaction encryption inside the non-async Viem code, and push it all into the provider code? However, last time I tried that it had trouble as it wasn't always going through the provider and I had trouble intercepting calls to the provider.

[CedarMist]

Lets document the reason for having the interval timer.

It be changed to setInterval, so it's continuous.

And the interval ID should be unref'd

[aefhm]

Agree. unref seems best right now.

[CedarMist]

Have you been able to reproduce the problem with a test? Would be good to have this in CI

[matevz]

Closing this in favor of #383