Johannes Rudolph
Johannes Rudolph
0-RTT doesn't even seem to be implemented in the JVM: https://bugs.openjdk.org/browse/JDK-8049402
https://github.com/akka/akka/runs/6361613518?check_suite_focus=true#step:5:3108
https://github.com/akka/akka/runs/6975926476?check_suite_focus=true
> I think we discussed it during the Scala 3 work and said the implicit extension wasn't a very important feature and that we could skip that for Scala 3...
A few ideas: * set `controller.kubernetes.io/pod-deletion-cost` for the oldest node, so it stays alive as long as possible (requires k8s api access) * Can we use `appVersion` for determining oldest...
There might be other k8s policies affecting the upgrade ordering as well, e.g. spread rules that maintain how a deployment is distributed to different AZs inside a cluster. We should...
I tried to see if we could derive something from k8s. There's a `deployment.kubernetes.io/revision` annotation that looks useful on the deployment and the replicaset but it's not easily accessible from...
@drewhk on gitter: > it has a breadth, though... > that complicates things > so maybe better to have a FlattenConcat specialized for this? > I.e. adding the optimization for...
FTR, I think the original behavior was the intended one and should not have been changed. By definition an InputBuffer is in front of a stream component and not after....
> `Source.fromPublisher` there is a buffer "in front"/inside/between the stream and the publisher making it feel like weird ergonomics Thinking about it again, I can see how you can see...