Can we skip polling when using broadcastTx ?

[williamchong]

Currently broadcastTx always polls tx for block inclusion. While timeoutMs and pollIntervalMs is configurable, setting them to 0 would make the function instantly throws a timeout error as a side effect.
Allowing polling to be skipped in broadcastTx() without throwing an error would allow batch sending of many transactions to be easier.

(I understand many messages can be included in one transaction, however in some cases we would want to send batch transactions with different memos)

[webmaster128]

Tracking the block inclusion is an essential feature of broadcastTx. If you don't need or want that feature, it buils down to just this line:

const broadcasted = await this.forceGetTmClient().broadcastTxSync({ tx });

which you can call directly if you have a Tendermint34Client at hand. You can do that manually.

An alternative approach to using Tendermint34Client is keeping the polling but not waiting for the promise resolutions. In https://github.com/noislabs/nois-bot/blob/74a5f9c8fc64c325f0d078420169a8514e3c3dbf/index.js#L205-L222 you find a multi-cast implementation where 3 broadcastTx to different RPCs are made in parallel. Once everything is sent, the code checks all the results.

[williamchong]

it's actually hard to get Tendermint34Client from stargateSigningClient since it's a private getter
To make usage easier, do you think skipping polling if timeoutMs is set to special value 0 or -1 makes sense? If yes I can try to make a PR

[webmaster128]

What if we add a broadcastTxSync next to broadcastTx which does not track the block inclusion? It would only contain those lines and then return after checkTx:

cosmjs/packages/stargate/src/stargateclient.ts

Lines 473 to 479 in e89a914

	const broadcasted = await this.forceGetTmClient().broadcastTxSync({ tx });
	if (broadcasted.code) {
	return Promise.reject(
	new BroadcastTxError(broadcasted.code, broadcasted.codespace ?? "", broadcasted.log),
	);
	}
	const transactionId = toHex(broadcasted.hash).toUpperCase();

[williamchong]

Yes that would work well 👍

[webmaster128]

@williamchong Can you have a look at #1396 to see if it solves your problem?

[williamchong]

@williamchong Can you have a look at #1396 to see if it solves your problem?

Yes it would, thanks

[webmaster128]

Done in #1396