Tatu Saloranta

No plans to implement that I know of; but wanted to add a note that such support would be tricky to do just for JSON; a lot of whitespace-skipping code...

True, implementation only considers properties found. It might be possible to figure out a fallback case, if (but only if) such could be uniquely determined. What could help here would...

I must admit that since this feature came as a contribution, I am not as intimately familiar with it as many other parts of the codebase, but it makes sense....

@drekbour makes sense -- I was to (but forgot) point out that reliance on intermediate parent types seems bit iffy to me as well, from modelling perspective.

I think this is implied already, but my understanding is that the deduction here relies on "discriminators", values that are distinct to a type; and that these should be relatively...

Leaving this here, in case anyone finds inspiration to further improve handling, but after improvements to `defaultImpl` handling (see #3055) there are no specific plans for improvement on this.

Thank you @lderienzo, that is an interesting way to describe it -- and Java (and thereby Jackson) definitely are more along more-static typing continuum. This does not mean that more...

@lderienzo Sure, that sounds great! I am not going to pretend this wasn't a rather complicated area, and with pieces that were not designed for extensibility. At the same time,...

While not optimal, tickets are fine, although I'd probably create a separate new one for implementation discussion (either here, or `jackson-future-ideas`). There is the gitter chat: https://gitter.im/FasterXML/jackson-databind which I try...

@ben-spiller Thank you for reporting this issue: it is indeed unfortunate. I wish I had paid closer attention to the contribution to consider backwards-compatibility aspect in more detail. While there...