Improved Key Estimation from Partitioned Indexes

c-treeACE SQL utilizes an estimated number of keys between target values as a measure of index selectivity during query optimization. When estimating key values between two target values, an excessive bias in key distribution could be encountered with partitioned files. This could result in an inefficient index choice for a query resulting in poor performance. The EstimateKeySpan() and KeyAtPercentile() routines has been improved to provide accurate key estimations for partitioned key values.