gistfile1.php

  function getKeys() {
    $params = array('props'=>'false','keys'=>'true');
    $url = RiakUtils::buildRestPath($this->client, $this, NULL, NULL, $params);
    $response = RiakUtils::httpRequest('GET', $url);

    # Use a RiakObject to interpret the response, we are just interested in the value.
    $obj = new RiakObject($this->client, $this, NULL);
    $obj->populate($response, array(200));
    if (!$obj->exists()) {
      throw Exception("Error getting bucket properties.");
    }
    
    $keys = $obj->getData();
    return array_map("urldecode",$keys["keys"]);
    
  }

This method sure is expensive.

200 keys
$bucket->getKeys() * 10

---
549ms
556ms
542ms
538ms
539ms
545ms
539ms
540ms
537ms
542ms

---
Total: 5427ms
Average: 543ms

keys=true and keys=stream are about the same.

Which is why it's not recommended...though I'm just using it on an admin panel until i start building indexes.

Just ran a map/reduce (no indexing) and it took even longer

$client->add($bucket->name)
    ->map("function(v){return [v.key];}")
    ->run();

---
645ms
620ms
621ms
626ms
623ms
628ms
627ms
622ms
625ms
628ms

---
Total: 6267ms
Average: 627ms

Not familiar with indexing in Riak.
I'd be interested to see how you do it.
I'm usually found idling in #riak