query optimizer fails to use index when sort spec is inconsistent with index spec

Beginning planning, options = INCLUDE_COLLSCAN 

Canonical query:

ns=test.server11568 limit=0 skip=0

Tree: $and

Sort: { a: true }

Proj: {}

=============================

idx 0 is { _id: 1 }

idx 1 is { a: 1.0 } sparse

Finding relevant indices

rated tree

$and

Planner: outputted 0 indexed solutions.

Planner: outputting a collscan:

SORT

---pattern = { a: true }

---fetched = 1

---sortedByDiskLoc = 0

---getSort = []

Child:

------COLLSCAN

---------ns = test.server11568

--------- filter = $and

---------fetched = 1

---------sortedByDiskLoc = 0

---------getSort = []

not caching runner but returning 2 results

> db.example.ensureIndex({title: true}, {sparse: true})

> db.example.insert({name: "Bob", title: "doctor"})

> db.example.insert({name: "Jim"})

> db.example.find().sort({title: true})

{ "_id" : ObjectId("5278370b13b0e338b3511f67"), "name" : "Bob", "title" : "doctor" }

> db.example.find().sort({title: 1})

{ "_id" : ObjectId("5278370b13b0e338b3511f67"), "name" : "Bob", "title" : "doctor" }

>

 db.example.ensureIndex({title: true}, {sparse: true})

> db.example.insert({name: "Bob", title: "doctor"})

> db.example.insert({name: "Jim"})

> db.example.find().sort({title: true})

{ "_id" : ObjectId("527836b596870941233e3da8"), "name" : "Bob", "title" : "doctor" }

> db.example.find().sort({title: 1})

{ "_id" : ObjectId("527836bb96870941233e3da9"), "name" : "Jim" }

{ "_id" : ObjectId("527836b596870941233e3da8"), "name" : "Bob", "title" : "doctor" }

>