Tatu Saloranta
Tatu Saloranta
Ok yes, recursion to re-order, since it's effectively linked list. Hmmh. I'll have to think about, no obvious better way to go about that.
I'm not happy with recursion due to possible stack exhaustion. But adding an extra data structure not great either. I think we need to have this as opt-in (`DeserializationFeature`) for...
One other note: while retaining order sounds like an improvement, it also has some backwards-compatibility aspects so I think that actual change should probably go against `3.x` branch, not `2.x`....
> I guess that makes sense, but it requires that I upgrade to v3 first, so it'll take a while... Understood.
This is sort of known, and due to implementation, where it is rather difficult to determine if property is or is not known. But obviously it'd be good to catch,...
@sgri this is an open issue (or, "feature"), exists in all versions. Will not be resolved before Jackson 3.0 since there isn't enough information available to do checks at level...
The main problem right now is that there simply is no knowledge of what "unknown" property might be, given the way deserialization of Unwrapped values is handled -- contents of...
NOTE: was not "figure out" for 3.0, still an architectural challenge.
Phew! Finally fixed thanks to @JacksonJang's contribution (#5504)! Will be in 3.1.0.
Oddly does not seem to speed up `jackson-benchmarks` CBOR test (in fact, ~5% slow down). Maybe `new String(byte[], int, int, Charset)` not more optimal or... need to do some profiling...