Support @JsonFormat(shape=STRING) for serialization of Range<T>

[chris-dekker]

I find the standard serialization JSON, while flexible, of Range<T> to be quite verbose:

{
  "lower_endpoint": 12,
  "lower_bound_type": "CLOSED",
  "upper_endpoint": 34,
  "upper_bound_type": "OPEN"
}

When manually processing the results, I prefer to shorter notation produced by, among others, the default toString() invocation: [12..34).

Similar to Instant, where we can use @JsonFormat(shape = ..) to control (de)serialization to use the ISO format or unix milis, it would be great if we could control the shape of the range this way. Keep the default as-is and only when @JsonFormat(shape = STRING) is provided, use the shorter notation.

Granted, this will only work for endpoints having scalar values, but that seems to cover the majority of use cases I've seen so far.

[cowtowncoder]

Sounds like a good idea for someone with time & interest to work on, and solid improvement to usability. Tagged as such.

[saydar31]

Assign it to me please,
Also I have a question: how should range without lower or upper bound look like?
(,1] or (-INF, 1], in my humble opinion second one looks better

[cowtowncoder]

@saydar31 I let you decide. But does Guava's Range.toString() offer guidance here?

[chris-dekker]

@saydar31 great to hear you're picking this up! Would prefer to stay consistent with the Range.toString() method indeed, which uses the ∞ symbol (\u221e)

[JooHyukKim]

@saydar31 great to hear you're picking this up! Would prefer to stay consistent with the Range.toString() method indeed, which uses the ∞ symbol (\u221e)

@chris-dekker now that you mention it, this needs to go on the doc. I will add notes to the PR that whatever symbol Jackson returns, the returned unicode of symbol would also be documented.

[cowtowncoder]

Merged for inclusion in 2.17.0