So as I am looking at the `GsonFeature` and trying...

# ktor

[trevjones]

So as I am looking at the GsonFeature and trying to implement something similar for jackson I am remembering why I did what I had done previously. I want to be able to write the json to the response socket as I am serializing it so that the response can begin sooner and I can potentially have less sitting in memory. Is there any existing or planned functionality for the pipelines to support streaming flows?

Context on why I want this. I previously built a database syncing system for android that interacts with very normalized data classes. The primary bottle neck was obviously the network but even then the latency before we even began to download a response was the bulk of it. Basically it was: query db, marshal all data from cursor into memory, build response envelope, serialize response, write response to socket. where each step would be slower with an increasing page size.