TODO
====

BSON
----
+ on insert/update: reject keys that start with "$" or "."
+ data support for common mongo "$symbols"
+ convert from/to json
+ tie in with native regex like python does?
  - on outgoing uncompile?
  - on incoming automatically compile
+ more time convertibles
+ map operations for BsonDoc (or should it be applicative?)

MongoDB
-------
+ support full level 0
  - handle query errors
  - hint, explain, $where
  - database profiling: set/get profiling level, get profiling info
  - pair mode connection
  - operations on database objects
    * getName
    * getCollection
    * add_son_manipulators
    * dereference (dbref)
    * error
    * eval
    * last_status
    * logout
    * reset_error_history
    * eval
  - database admin
    * getAdmin
    * getProfilingLevel
    * setProfilingLevel
    * getProfilingInfo
  - collection
    * modify
    * replace
    * repsert
  - index operations
    * ensureIndex / index existance caching
  - misc operations
    * explain
    * getOptions
    * getName
    * close
    * group
    * distinct
    * map reduce
  - cursor object
    * hasMore

  orderby (sort)

  optional:
  - automatic reconnection
  - buffer pooling
  - advanced connection management (master-server, replica pair, Option_SlaveOk)
  - Tailable cursor support
+ support safe operations
+ auto-reconnection
+ auto-destoy connection (how?/when?)
+ pymongo returns the new ObjectId(s) on insert
+ don't read into cursor until needed, but have cursor send getMore before
  it is actually out of docs (so network is finished by the time we're ready
  to consume more)
+ support a LIMITed quickFind

Misc
----
+ learn more about haskelldb, anything we can learn from there
+ go through pymongo api and figure out what parts to adopt (also look
  at other languages?)
  - database_names()
  - collection_names()
+ support for aggricated commands like listing collections
+ kill prefix on data types "eg QO_*"?
+ javascript
+ tailable cursor support
  - only close cursor when cursorID is 0
  - have to create loop that sleeps and retries
  - lazy list support
+ error handling?
+ concurrency (share connection?)
+ is there a garbage collector hook that will let us free cursors and connections?

Tests?
Documentation
  - ref
  - tutorial

GridFS


pretty printer
deep "lookup" function (other deep Map functions?)
how to make bytestrings less painful
custom Show/Read instance that looks more like json
make sure NULLs aren't in created table names