Even if you abstract away the indexes, when working with datasets that large you have to worry about whether the standard implementation makes the same class of

Even if you abstract away the indexes, when working with datasets that large you have to worry about whether the standard implementation makes the same class of error. For example, using Java's binarySearch on arrays of length over 2^30 was broken until 2006[0][1]. [0]: http://googleresearch.blogspot.com/2006/06/extra-extra-read-all-about-it-nearly.html http://googleresearch.blogspot.com/2006/06/extra-extra-read-... [1]: http://bugs.java.com/bugdatabase/view_bug.do?bug_id=5045582 http://bugs.java.com/bugdatabase/view_bug.do?bug_id=5045582