Deserialize ArrayItems and JsonValueFormat correctly

[georgewfraser]

No description provided.

[cowtowncoder]

I assume it is legal to use array, or a single item here?
This works, I can probably simplify it after merge a bit.

[cowtowncoder]

I think I'd like to know bit more about what exactly is being fixed and why. I can see that handling of JsonValueFormat is changed, but not sure why so: default Enum serializer would use upper-case name(), new code would expect to use lower-case desc. Is that something JSON Schema specification mandates?

If the change is needed, there are also ways to change definition of JsonValueFormat, which is bit simpler: for example, adding @JsonValue to toString() method. This seems preferable, as it should support both serialization and deserialization.

Handling of item array makes sense, but it would be good to have unit test(s) to show how things work better now.

[georgewfraser]

JsonValueFormat should be serializing and deserializing using the lower-snake-case values. These values are part of the JSON schema spec. If you run this example https://gist.github.com/georgewfraser/196c84c5072a800e9cb9 through a JSON schema validator, for example http://jsonschemalint.com/draft4/#, you will see that "date-time" is a real format that gets validated, but DATE_TIME is meaningless.

I added some tests, and fixed some existing tests.

[cowtowncoder]

And just to make sure, lower-snake-case is what JSON Schema v3 defines? (since that's what module supports). I know v4 has changed many things, and this has caused lots of confusion.

[georgewfraser]

Yes, all versions use lower-snake-case.

[cowtowncoder]

Ok thanks. Just wanted to verify.

[cowtowncoder]

Ok, merged, will do some minor cleanup.

One thing I did not, wrt JsonFormatValues, as per:

http://json-schema.org/latest/json-schema-validation.html#anchor104

that existing list does seem to fully align with that; and (for example), whereas we have "host-name", spec seems to have "hostname". Similarly with "ip-address" vs "ipv4". This could be some difference between v3 of json schema vs later verisons.

I guess I won't be changing that quite yet, but thought I'd note it.
Also, use of enum for format values does seem bit wrong, wish I had known to bring it up when functionality was first merged (whole JSON Schema handling is contribution, I did not write it).

[georgewfraser]

What is your plan for moving to v4? I'm fixing some v3/v4 differences on https://github.com/fivetran/jackson-module-jsonSchema

On Oct 7, 2015, at 10:58 AM, Tatu Saloranta notifications@github.com wrote:

Ok, merged, will do some minor cleanup.

One thing I did not, wrt JsonFormatValues, as per:

http://json-schema.org/latest/json-schema-validation.html#anchor104

that existing list does seem to fully align with that; and (for example), whereas we have "host-name", spec seems to have "hostname". Similarly with "ip-address" vs "ipv4". This could be some difference between v3 of json schema vs later verisons.

I guess I won't be changing that quite yet, but thought I'd note it.
Also, use of enum for format values does seem bit wrong, wish I had known to bring it up when functionality was first merged (whole JSON Schema handling is contribution, I did not write it).

—
Reply to this email directly or view it on GitHub.

[cowtowncoder]

No solid plan. Some discussion on dev list (and with some developers). I think it is easiest to do via new module, instead of trying to add modality; similar to how Hibernate module (v3, v4, v5) works.
One open question there would be whether to do Maven sub-projects/modules, or separate git repos.
I am guessing separate repos would be better eventually, considering that v3 should eventually be deprecated.

It's probably best to discuss this further on jackson-dev list; not all/many developers read issue updates on regular basis.