JsonNode? missing value should be deserialized as null and not as NullNode

[novtor]

Issue: #490

[dinomite]

This looks good. Could you rebase this against the 2.13 branch? An issue like this isn't going to trigger new 2.11 releases and I don't think @cowtowncoder is planning to do any more 2.12 releases.

Also, add an entry to the Contributors: section of CREDITS-2.x and send a signed CLA to info at fasterxml dot com, then I can get this merged.

[dinomite]

Ugh, I meant to push to your branch but accidentally pushed to the 2.11; would you re-open this PR?

Also, I have a commit that I couldn't push, if you'd apply this patch and commit it, please.

491.patch.txt

[cowtowncoder]

@dinomite One very quick note: Jackson-databind 2.13.0 (post rc2) now has new method:

JsonDeserializer.getAbsentValue()

used to solve issue:

FasterXML/jackson-databind#3214

in which missing (absent) value passed via @JsonCreator annotated constructor became NullNode instead of plain null.

As such this method might make more sense to call than getNullValue() -- semantics differ at least for JsonNode.
In future we'll have to think of whether to similarly change behavior for "referential" types like Java 8 Optional.
But I think semantics here would be for "absent" value more so than "null replacement" (which mostly/only makes sense for primitive values for which null is illegal to pass due to unboxing).

[dinomite]

@novtor I both accidentally merge-closed this and forgot to mention you—would you reopen this PR and add that patch?

[novtor]

@dinomite, here is the PR for 2.13: #500

[cowtowncoder]

CLA received for @novtor.