Uwe Schaefer comments

Results 89 comments of


                                            Uwe Schaefer

Projection::postprocess recieves a list that should be a collection

Picking up your fair point about `List foo(List)`, wouldn't that be even more confusing? "Then in most cases you could just override the second one in the projection:" most of...

Projection::postprocess recieves a list that should be a collection

"I’m also wondering how factspecs and post filtering (like checking userDeleted.userId == aggUserId in the handler for UserDeleted) are connected. Sure one is server side for selection, the other client...

Projection::postprocess recieves a list that should be a collection

I think your former comment mostly deals with #3717 which is not connected to postprocess. Lets keep this issue about the postprocess method in the projection interface.

Performance Bug: Notify

"Looks like currently we send one notification per inserted row (right?), that could maybe be optimised in case of many concurrent or batched inserts." Nope, with batch inserts, we collapse...

Performance Bug: Notify

I don't get what a pre-commit ping is for. we only want to be notified of comitted changes. If ever, for some reason we miss a notification, that is not...

Performance Bug: Notify

Might be intersting https://redis.io/docs/latest/integrate/redis-data-integration/data-pipelines/prepare-dbs/postgresql/

Check if throughput can be increased if we only lock namespaces on inserts

this is a bit wasteful in terms of sql statements executed. Instead for aquiring lock by lock from java, we should call a function in the db to lock a...

Check if throughput can be increased if we only lock namespaces on inserts

Must be equivalent to "all known by now"?

Check if throughput can be increased if we only lock namespaces on inserts

bit hard to comprehend. maybe you could prototyp it?

Check if throughput can be increased if we only lock namespaces on inserts

I have a less complex, hopefully comparably performant approach in mind, that i'd like to test. Could you push your benchmarking code? (maybe on a sep. branch?)