Fixing `contains` and `minContains`
gregsdennis opened this issue · comments
Currently contains
and minContains
are defined so that if minContains
has a value of 0
, it changes the validation result of contains
. There are two problems with this:
- Keywords are allowed to change their own behavior, but not the behavior of other keywords
Keywords MAY modify their behavior based on the presence or absence of another keyword in the same schema object - Core 7.2
contains
is defined in Core, whereasminContains
(andmaxContains
) is defined in Validation (the spec, not the vocab).
Ideally, all of the logic should be defined for contains
: contains
looks at minContains
and maxContains
to determine the appropriate bounds. This behavior fixes the first problem.
In order to address the second, we'd either need to move contains
to Validation (but it's still an applicator), or we'd need to move minContains
and maxContains
to Core. There's also the issue that unevaluatedItems
now depends on contains
, which seems to support moving min/maxContains
.
So, do min/maxContains
work in Core? What vocab would they be a part of? The only vocab is applicators and they're not applicators.
Alternatively, can we move the entire applicator vocab out of Core into Validation? Does Core need applicators?
For some reason I though we had already taken care of this, but if we haven't, we definitely need to. I think it makes perfect sense for min/maxContains
to be in the applicator vocabulary. The group of contains
/minContains
/maxContains
are really one unit and minContains
and maxContains
are just modifiers of contains
, so contains
should determine where the whole set should go, which means it belongs in applicators.
Okay, yeah, it looks like they're grouped under "Other Keywords for Applying Subschemas." I didn't know we had done that already. (Although there is a typo of minContians
in there 😄)
Closing this as the specific topic is resolved (already done).