Henry Andrews

@jdesrosiers I think I may have checked out the wrong branch or something when looking at this and thought there was a lot less referencing 3987 than there is (I...

@jdesrosiers I think the only place I would update would be the reference for the normalization process, as RFC 3987 §5 has numerous subsections with a good bit of IRI-specific...

@jdesrosiers `contentSchema` is specifically ***not*** an applicator or schema location keyword (meaning a keyword like `$defs` that primarily serves as a reference target). Its value is returned as an annotation,...

What would it even mean to treat it like a normal schema? How would that solve the problem at hand, which is using it as an annotation? Can you explain...

I realize the above may be phrased in a more demanding way than I really intended. I am honestly curious as to whether you see something I am missing. When...

The problem is that this does not handle the case where the schema containing `contentSchema` is not directly accessible to the caller. For example, if the schema evaluation happens on...

> if there is some real(-ish) example of when this could possibility be the case There's one buried somewhere in the `unevaluatedProperties` but I'm not slogging through that mess to...

@jdesrosiers I definitely see the points you're making, and I agree that having the complete containing schema resource (and any information needed to fully resolve its base URI if it...

@jdesrosiers > It wouldn't surprise me if you were the one person who has has ever tried to do something with this feature. I have never added a keyword purely...

@notEthan could you move the schema vs data stuff to #1307, please.