sroze / fossil Goto Github PK

The SSE stream gives the consumer as much data as it can get. It will give it hundreds of megabytes per seconds if need to be. However, 1) the client can't handle this and 2) it's not very smart to read all of this at a time from the database perspective.

We need to be able to handle reading large SSE stream without impacting the performance of Fossil too much. Using Postgres' cursors is probably the way forward (example with pgx). Wikimedia's KafkaSSE had a similar issue.