Support of transactional partitoned source using multiple transactional sinks

Hi,

I’ve found a rather interesting piece of code https://github.com/akka/alpakka-kafka/blob/main/tests/src/test/scala/docs/scaladsl/TransactionsExample.scala#L145.

What happened to the Transactional.partitionedSource? Will this come back at some point in the future? What is the current alternative if you want to increase transactional throughput? Scaling the application horizontally, clamping the transactional sources to specific partitions and using different transactional ids?

We did a number of experiments with the Transactional.partitionedSource a few years ago but were not able to make it work reliably. We currently don’t have plans to pick it up again.

Cheers,
Enno.