Serde takes up ~5-10ms of time in some calls, particularly for large objects. Switching to a perf-focused serde serializer would be a non-negligible benefit, at the cost of less transparency / adaptability for the transport protocol.
I think it's worthwhile. As long as the serialization is standard across other languages, it's not a particular barrier to other PDKs. And it's a fairly trivial improvement.
Serde takes up ~5-10ms of time in some calls, particularly for large objects. Switching to a perf-focused serde serializer would be a non-negligible benefit, at the cost of less transparency / adaptability for the transport protocol.
I think it's worthwhile. As long as the serialization is standard across other languages, it's not a particular barrier to other PDKs. And it's a fairly trivial improvement.