Roadmap

Notes

Ideas thrown in the wild. Not necessarily in order.

For 1.2.x

Simplify pre-validation processing when possible

Some schemas don't have JSON References. Some schemas can be fully validated before instance validation.

Account for these, simplify pre-validation.

Get rid of custom unchecked exceptions

They don't really serve a purpose after all. The JDK has plenty of them already.

Generalize use of msg-simple (DONE)

Using enums is a pain, and not extensible. Allow for extensibility as well (this one should be easy).

Move JSON Patch and JSON Pointer out of this package (DONE)

Both are in -core for now, but can be used by other packages.

Path redirects

Right now, only schema redirects exist. Also implement path redirects.

Make schema loading use processors

Right now, all the logic is "hardcoded" within SchemaLoader. Split into different processors. Particularly beneficial for URI rework before loading.

Draft v4 Hyperschema validation support (DONE)

Use uri-template now that it is written and functional.

Make schema walking worth it; what about instance walking?

Schema walking is tested, fine; but it is not really put to use.

Also, what about the ability to walk instances? What use would there be?

Future plans

Some of these may, or may not, make it in 1.2.x.

Write examples

Not easy to think of "useful" examples for such a generic library...

Combining outputs into a new input

Right now, this is done by hand. Scenario:

• processor p1 has output O1;
• processor p2 has output O2;
• processor p has an input consisting of a combination of O1 and O2.

Find a generic way to allow such combinations. Find a generic way to "wait for" p1 and p2 before invoking p.

Dependency injection?

The chain of builders to generate configurations etc is quite long. See if dependency injection can simplify matters somewhat.

Draft v3 hyperschema validation support?

Very complicated. And probably not worth it.