Clarification - Purpose of "mergeEvaluated" in CodeGen?

Hello! In implementing some custom keywords, I’m not sure of when/why to use the mergeEvaluated method of KeywordCxt class, and was wondering if I could get some clarification? So far, I haven’t been able to find anything on it within the docs or the src.

Presumably it merges a subschema’s SchemaCxt into the current keyword’s KeywordCxt, and I see it’s employed by many existing keywords already, but I’m not sure what effect that actually has. When I’ve overwritten standard keywords to omit their calls to mergeEvaluated or mergeValidEvaluated, it never seems to change the behavior of my validation or its output. Any explanation would be appreciated. Thanks!

[epoberezkin]

It statically (that is, during compilation time), when it can, or dynamically (during validation time), when it can’t do it statically, collects object properties and array items that were evaluated - this information is then used by the keywords unevaluatedProperties and unevaluatedItems to determine which properties/items they should be applied to.

If you do not use unevaluatedProperties / unevaluatedItems keywords that would have indeed no effect, but it’s impossible to determine during the compilation whether these keywords are used, as they can be in another schema, so as long as these keywords might be used this information has to be collected.

[epoberezkin]

the best place would be here: https://ajv.js.org/components.html or here: https://ajv.js.org/codegen.html