Hi,

like the title sais. Is there a rewind() or reset() method? If not, could you move it to suggestions? :)

Which approach would be less expensive when I need to iterate over a result n times?

a)

var cursor = Spark.runtimeCollection("col").find(query);
var doc;

while(cursor.hasNext())
{
    doc = cursor.next();
    // do something with doc
}

// get new cursor
cursor = Spark.runtimeCollection("col").find(query);

while(cursor.hasNext())
{
    doc = cursor.next();
    // do some other stuff with doc
}

 b)

var result = Spark.runtimeCollection("col").find(query).toArray();
var doc;

for(var i = 0, n = result.length; i < n; i++)
{
    doc = result[i];
    // do something with doc
}

for(var i = 0, n = result.length; i < n; i++)
{
    doc = result[i];
    // do some other stuff with doc
}

In the first approach I would save the document (if modified) to have changes made in the first iteration available in the second iteration.

In the second approach I would iterate one useless time on fetch over the result. Also the object could be huge. But I have less database calls (no saving in between needed, because the object reference persists).

A compromise is to make the first iteration with the SparkMongoCursor and push everything to an array for later usage. But this is not very usefull in many cases.

Also in my example I used 2 iterations, but the actual question is which version performs better with n iterations.

Would be interesting to know how other people handle this.

Cheers

David