Possible bug/enhancement for Hyperblock mode

[alexberce]

Hi there,
We are using this library to process the transactions sent to some of our smart contracts.

When using the BlastApi gateway we get a lot of errors (error code 500) and from the discussion that I've had with the BwareLabs team, this usually means that we requested a block from the future (not finalized yet).

After a quick look at the library code, I noticed that it uses erd_nonce for the current nonce, and not erd_highest_final_nonce. Shouldn't erd_highest_final_nonce give us the last final block?

private async getCurrentNonce(shardId: number): Promise<number> {
    const shardInfo = await this.gatewayGet(`network/status/${shardId}`);
    return shardInfo.status.erd_nonce;
}

I'm proposing the following non-breaking changes to the library:

1. Add a new optional property that allows us to specify which nonce to use

export enum TransactionProcessorNonceStrategy {
  CurrentNonce = 'erd_nonce',
  HighestFinalNonce = 'erd_highest_final_nonce',
}

export class TransactionProcessorOptions {
  ...
  nonceStrategy?: TransactionProcessorNonceStrategy = TransactionProcessorNonceStrategy.CurrentNonce
}

2. Update the getCurrentNonce code to use the the new property

private async getCurrentNonce(shardId: number): Promise<number> {
    const shardInfo = await this.gatewayGet(`network/status/${shardId}`);
    return shardInfo.status[this.options.nonceStrategy];
}

I can create a PR if you are willing to include it in this library.

[tanghel]

hello!

this sounds like a great idea!

it would indeed create some extra api calls, but then we would have the guarantee that our calls don't fail because of "nonces in the future" error

another approach I was thinking was to perform the requests per shard every 6 seconds (after maybe performing an initial calibration) thus greatly reducing the possibility for the "nonces" in the future error.

Or maybe an approach where we integrate both functionalities, maybe in 2 steps. by performing one request every 6 seconds we can "afford" to make one extra request to get the current nonce to make sure we can perform the fetch

[alexberce]

I'm not sure that I understand how performing per shard requests would help.
We are using the library in Hyperblock mode, which should not be concerned with the individual nonces of each shard.

After making a request to /network/status/4294967295, we (the library) received a "erd_nonce" value of 14758924 but the request to /hyperblock/by-nonce/14758924 resulted in a 500 error.

My proposal allows us to stay one block behind, eliminating the "nonces" in the future error completely and does not introduce any additional API calls.