I am about halfway through a new implementation that only has the the "token"-level parsing (I think you call that "push"). It seems to simplify the design, although a lot more code for token checks, but I think it is the right thing to do. In this, it doesn't seem to hard to support both Map formats, basically by key type. If key type is plain (String, number, date and few others) then the object format will be used and otherwise the key/value entry format is used.
As for "_type", I am not sure yet, but I think ParameterizedTypes will be required and inability to provide that will simply serialize with Java Object Serialization (if it cand) or fail. I will report back when I am nearer to the tricky bits. Niclas On Sun, Jul 3, 2016 at 9:53 PM, Paul Merlin <[email protected]> wrote: > Niclas Hedhman a écrit : > > Paul, > > What is the difference between the two Map deserialization methods in > > ValueDeserializationAdapter? > > > > One is parsing using tokens, and the other seems to be using JsonNode. > One > > is has keyDeserializer in the arguments and the other don't use that. I > am > > very confused... > The whole deserialization supports both push and pull parsing and > dynamically switch from the former to the later when in eager need for > _type. Both parsing modes use the same Map format. > > -- Niclas Hedhman, Software Developer http://zest.apache.org - New Energy for Java
