mapreduce - Aggregating heterogeneous documents in MongoDB -

i writing tool inspect mongodb session store, in structure namespaces containing variables containing information on variables, roughly:

{     "some namespace" : {         "some variable" : {             "t" : "serialization type",             "b" : "data value",             "c" : "java class type"         },         "some other variable in namespace" : {             "t" : "serialization type",             "b" : "data value",             "c" : "java class type"         },         ...     },     ... // more namespaces } 

the names of namespaces , variables different. wrote map-reduce job loop through namespaces , loop through variables in each namespace , emit each variable in each namespace.

however, want try switch mongo's aggregation framework avoid potential write lock problems. based on docs, seems have specify names of variables $group on, doesn't seem compatible heterogeneous data mine. there way involve iteration in aggregation invocation?

currently, use aggregation framework, have know field names write query. there some proposals remove limitation.

---

right now, given design, use aggregation framework, have consolidate @ regular intervals data using map-reduce more or less that:

{     namespaces: [       { name: "some namespace",         variables: [           { name: "some variable",             t : "serialization type",             b : "data value",             c : "java class type"           },           { name: "some other variable in namespace",             t : "serialization type",             b : "data value",             c : "java class type"           },           ... // more variables         ]       },       ... // more namespaces     ] } 

now relatively easy query using aggregation framework. admitting may work on stale data...