I was just tracking some slow queries and found queries like:

db.Example.find({ "indices.text":

{ $all: [ "le", "mois", "suivant" ] }

, lc: "fra", group: "all" })

giving

{
"cursor" : "BtreeCursor indices.text_1_lc_1_group_1",
"nscanned" : 244471,
"nscannedObjects" : 244471,
"n" : 34,
"millis" : 6211,
"nYields" : 6,
"nChunkSkips" : 0,
"isMultiKey" : true,
"indexOnly" : false,
"indexBounds" :

{ "indices.text" : [ [ "le", "le" ] ], "lc" : [ [ "fra", "fra" ] ], "group" : [ [ "all", "all" ] ] }

}

but

> db.Example.find({ "indices.text":

{ $all: [ "suivant", "le", "mois" ] }

, lc: "fra", group: "all" }).explain();
{
"cursor" : "BtreeCursor indices.text_1_lc_1_group_1",
"nscanned" : 1948,
"nscannedObjects" : 1948,
"n" : 34,
"millis" : 102,
"nYields" : 0,
"nChunkSkips" : 0,
"isMultiKey" : true,
"indexOnly" : false,
"indexBounds" :

{ "indices.text" : [ [ "suivant", "suivant" ] ], "lc" : [ [ "fra", "fra" ] ], "group" : [ [ "all", "all" ] ] }

}

I can manually re-order my inputs to the $all putting longest strings first, but perhaps Mongo should have similar logic built in since longer strings should naturally be less common and therefore need less scanning?