In order to "simplify" the JSON saved in the database and make it more compatible with other systems/tools/etc. we opted to store the JSON without a type root. In general this works well, but for value queries at the "root" the only way to correctly match is with a range index.
A container query would work if there was a root node there, and removes some of the "free-floating" value matches. Path range indexes could then be specified at a type level, and more general indexes could be created as element range indexes.
Add a type persistence strategy of "ROOT" that would tell the converter to include the root when it serializes.
In order to "simplify" the JSON saved in the database and make it more compatible with other systems/tools/etc. we opted to store the JSON without a type root. In general this works well, but for value queries at the "root" the only way to correctly match is with a range index.
A container query would work if there was a root node there, and removes some of the "free-floating" value matches. Path range indexes could then be specified at a type level, and more general indexes could be created as element range indexes.
Add a type persistence strategy of "ROOT" that would tell the converter to include the root when it serializes.