Fixes #1573

[senojj]

Please excuse the trailing whitespace removal. My IDE likes to keep things tidy.

Essentially, the logic was cutting short the processing of settable properties. I deleted the code which caused the premature return, and now all settable properties are able to be processed appropriately with the object construction happening at the end of the function after the primary loop.

[cowtowncoder]

Would it be possible to have a unit test for this? Part of this would be to ensure fix, but most importantly guard against regression. Also would be good to know if master has the issue: code has changed for this method and it's possible problem either has been fixed, or will have to tackled differently.

[senojj]

Absolutely. I apologize for not already providing tests for this. I was in a hurry, and just shoved everything I had up. I'll provide the tests sometime tonight or this weekend.

[cowtowncoder]

@jjware np at all, thank you for submitting this!

[senojj]

I will have tests published to this pull request shortly. I just wanted to mention that while building my tests, I found out that using a builder constructor parameter annotated with @JsonUnwrapped does not work and yields an exception pertaining to unknown fields. This makes me think that the parser is not picking up on the @JsonProperty annotations of the unwrapped class. This is a separate issue that I will try to address after finishing with this one.

[cowtowncoder]

@jjware Problems with unwrapping are not entirely unexpected, since no introspection is or can be done from that point: all code knows is that POJO in question contains some properties, and so anything otherwise unknown must be buffered, and attempt to bind them must be done at a later point. I really wish I had not implemented @JsonUnwrapped at all, it's a messy implementation.
But in all likelihood if it's to be supported, handling needs to be rewritten for 3.x to access and use information on contained properties: this allows detection of actually unknown properties.

In the meantime it sounds like initial problem is resolved so I'll merge this.
Thank you again!

[senojj]

@cowtowncoder I can completely understand how the @JsonUnwrapped implementation can get pretty messy, however, I will tell you that I absolutely love what you have done with it. It's existence has removed the need for a transformation layer between the target API models and and my domain models.

[cowtowncoder]

@jjware Thank you for sharing this. I do know that it is widely used (if not for nothing else due to bug reports :) ), and due to wide usage will do my best to try to keep it working. Thank you for helping with this.

For Jackson 3.x I think it should be possible to rewrite handling to work much better; and this could also help with some other related things, such as "any" properties.