Continued work over Distributed Data #2441
Conversation
|
Does this need to be dove-tailed with #2421 at all? |
|
No, those are custom serializers, but they use Hyperion internally. On the other side I'm thinking, if we should fall back to proto-buf here: current version assumes, that all elements for serialized collections will use Hyperion as well, which doesn't have to be true. We might potentially run into situation where we want to use custom serializers for internal elements representation. |
3211f3a to
5e33cdd
Compare
|
Ultimatelly using protobuf here is not a good option - the major point is the fact, that we have heavily generic collections here, causing a lot of pain when working with protobuf. Therefore for the time being I'm proposing to use existing serializer (a slightly optimized one Hyperion-based). This PR is ready for review and to merge. |
| @@ -63,7 +63,8 @@ public static StringBuilder AppendJoin<T>(this StringBuilder sb, string separato | |||
| index++; | |||
| sb.Append(separator); | |||
| // ReSharper disable CompareNonConstrainedGenericWithNull | |||
| if (enumerator.Current != null) | |||
| current = enumerator.Current; | |||
| if (current != null) | |||
|
@Horusiath looks good to me |
This PR brings few bug fixes, more API methods for distributed data collections and optimized, Hyperion-based serializers for ddata module (original Akka uses Protobuf).