sumo querying ideas

query.clj

; 2i queries
(index-query conn :age 25)
(index-query conn :age 21 80)
(index-query conn :last-name "a" "f")

; is it too much of an abstraction to launch
; a mapreduce query which gets the docs and not
; just the keys?
(index-query conn :email "foo@example.com" :return-body true)

; map reduce calls start with an input, then
; a vector of map or reduce funcs
(mapreduce conn (:index :age 21 80)
       [(:javascript "function (obj) {...}")
        (:built-in "reduce_sort")]))

; links
; starts with a bucket/key
; and then a list of bucket/link-tags
; specs
(link-walk conn "bucket" "key"
  [("other-bucket" "tag1" :keep true)
   ("second-bucket" "tag2")])

index-query seems fine, I think fetch-body is fine too since it's a common case. Does it need to accept a connection and bucket though? I haven't used 2i much, tbh.

For query, do you not need to distinguish between map and reduce in the API? Does Riak not care, I thought it'd need to know what each function was "for"? I think it'd be useful for code reading to do something like (map/js "func...") (map/builtin "foo") (reduce/builtin "bar"). I could be totally wrong. :)

Link walking looks good to me, too.

• re: connection argument; yes, I just forgot about it.
• re: naming phases; yep, another omission

@bretthoerner, what do you think about these options for identifying language/func-type?

; named funcs
(map/js "func")
(reduce/js "func")

; keywords
(:map-js "function () {...}")
(:reduce-builtin "reduce_sort")

; hash
{:lang :js :type :map :src "function () {...}"}

I created issue #9 for mapreduce