if i have a collection of documents like:

{ name: 'fred', children: [

{ name: 'bill', age: 10 }

,

{ name: 'bob', age: 20 }

,

{ name: 'ed', age: 30 }

] }

and i want to operate on all children whose age is greater than 10, i don't believe there is a direct way to accomplish this.

i'm coming at this from the context of this jira entry: https://jira.mongodb.org/browse/SERVER-3347

i believe in the multi-geo case the document is returned once for each hit, which implies that multiple elements of an array met the search criteria.

in the multi-non-geo case the document is returned only once, i think it would be best to remain consistent, and in SERVER-3347 i suggest returning the document only once (after all returning the same document multiple times doesn't add much value),
but i allude to a "wrapper" document containing "meta" information on which elements of the array met the uery criteria.

i think the same pattern may be applicable to the multi-non-geo case, in a way to address the case in between the $ position operator and the multi (apply to all) mechanism.

so if there was a way to return something like:

{
hits: [

{field: "children.age", index: 1 }

,

{field: "children.age", index: 2 }

],
document:

{ // the actual document }

}

i think it would give the caller enough information to operate on those elements of the array that were of interest...?