Difference with generateTransactionChunksAsync()

[elliotsayes]

There is a difference in the data_root generated between generateTransactionChunks() and generateTransactionChunksAsync() when the input data is a multiple of MAX_CHUNK_SIZE. This is because chunkData() in generateTransactionChunks() has an additional zero-length chunk at the end, which in turn influences the output of generateLeaves() => buildLayers() => generateProofs(). The chunks and proofs returned remain identical as they are cleaned up before returning:

// Discard the last chunk & proof if it's zero length.
const lastChunk = chunks.slice(-1)[0];
if (lastChunk.maxByteRange - lastChunk.minByteRange === 0) {
    chunks.splice(chunks.length - 1, 1);
    proofs.splice(proofs.length - 1, 1);
}

I'm not sure if this actually makes a difference in terms of the validity of the transaction, but maybe it is worth mentioning that the outputs of the two functions can sometimes be different.

See the failing test here: master...elliotsayes:arweave-stream-tx:chunk-size-multiple