>Gregory,
>
>>I always do. And why not ?
>
>Because an index on the column itself would be faster. Since bintoc() by default returns a 4 byte value and integer is a 4 byte data element the index tree will be the same size so why waste cpu cycles generating the bintoc conversion?
>
>bintoc() was designed to create smaller concatenated column indexes than we could prior to it being added to the language.

It's a habit. I've always stayed away from indexing on Integer, Date, DateTime, Numerics and doubles

If bintoc() sometimes, not always, fails on views and cursors and not on tables (yet), who says that an index on an integer won't