map/reduce to get a bounding box of geo-indexed docs

[Daniel W]

I have a collection with this index.

"key" : {
"g" : "2d",
"c" : 1,
"s.s" : 1
}


I'm trying to determine the bounding box for all docs with a certain
value of c, so that I can randomly sample from within that box.
I wrote a map/reduce in python, and when I run it, it just hangs for
hours, even on a query that returns only 2 documents. Is there a bug
in my map/reduce? Is there a better way to do this?

def boundingbox(campaign, collection):
map = Code("function () {"
" emit ('minlt', this.g[1]);"
" emit ('maxlt', this.g[1]);"
" emit ('minln', this.g[0]);"
" emit ('maxln', this.g[0]);"
"}")

reduce = Code("function (key, values) {"
" if (key=='maxlt' || key=='maxln') { "
" var thismax=-180;"
" for (var i = 0; i < values.length; i++) {"
" thismax = Math.max(thismax,values[i]);"
" }"
" return thismax;"
" }"
" else {"
" var thismin=180; "
" for (var i = 0; i < values.length; i++) {"
" thismin = Math.min(thismin,values[i]);"
" }"
" return thismin;"
" }"
"}"

)
resultcoll = collection.map_reduce(map, reduce, "myresults",
query={'c':campaign})
results=resultcoll.find()
results = [row for row in results]
return results

[Spencer T Brody]

At first glance that seems correct to me.  Take a look at db.currentOp() to see what's happening while the MR is running, and also look at the mongod logs and see if you see any warnings or error messages.