* The classical problem when case class is extending a marker interface that is
configured to a specific serializer, java.io.Serializable and the marker
interface have no order. Multiple serializers found and then it is picking
the first one.
* This changes that behavior so that if multiple serializers are found it gives
JavaSerializer low priority and chooses the other serializer.
* This might look like a scary change from a compatibility perspective, but
it should be fine becuse it is not influencing deserialization. The previous
choice for serialization was pretty random so this choice should not make
it worse.
AFAICT there was nothing ensuring the order of messages when sent to the
shard and the region so first checkthat the passivation has happened
before sending another add in the test
Refs #24013
Giving more clarity on which whitesource project/product name is used
and better output when a violation is found.
I noticed whitesource errors during our pr validation, this might help
pinpointing what's going on there.
* Do not try to submaterialize inner source when outer stream already cancelled #23656
* Use StreamDetachedException
* Sleep a little
* Updated exception text
* +str Add throttle combinator without bucket size in parameters
* Add mima filter plus comment formatting
* Reviwed throttle messages across all classes
* move mima filter to 2.4.6
* change to throttleEven, plus grammar fixes
* fix formatting
* changed mima filter according to new API
* fixes accoring to feedback
* Handle race where broadcasthub consumer already cancelled #23205
* Don't use the same thread ec, import cleanup
* Complete feedback promise when async event has been executed #23953
* No need for double-deregister-guard after invokeWithFeedback was fixed.
* Test coverage for async callbacks #23953
* Updated docs on AsyncCallback #23953
* MiMa
* Completing a promise of scaladoc
* Improved test coverage and handling of async invokation failing or throwing
* MiMa
* Review updates
* test was not always using serialization because it was just one
entity which could be local or remote
* the test serializer was wrong
* multiple serializers found so was using java serializer
This once failed where the intial shopping cart before adding an item
was not empty. I've also changed the item prices to be different for
each test just in case we're picking up state from a previous test
(which seems unlikey given the pid is a random uuid)
