Closed chickenlj closed 3 years ago
@chickenlj Could u pls. describe it with more details?
Firstly, this is a performance tuning task. Dubbo internally have a name-id mapping for each serialization implementation, for example,
What I suggest is, use serialization id instead of name whenever we need to negotiate serialization between Provider and Consumer.
At present, I think it's no big deal whether we are using serialization name or id. Considering the tasks we have on hand, we can tune it later.
When user set
serialization="hessian2"
, use corresponding id internally.