The iHub Wish List

[Bob Zumbrunnen]

It would be possible, but since it involves the most expensive part of the search, it would at least double the expense and the time it'd take to return results. So I had to make then case-insensitive.

Keep in mind, too, that there are aspects of search that're counter-intuitive. For example, narrowing down a date range or authors or boards does nothing for query execution times. No matter what those criteria are set to, the whole message table is first searched for occurrences of the entered words. Then the results of that search are narrowed down according to the other criteria.

I wish it would/could work the other way around, but it doesn't. The only way it could would be a home-grown indexing scheme, which is still a likelihood in the distant future.

Using the Microsoft Search Service (which SQL Server does), no matter how you write the code or the order you put the clauses, the entire messagebase index is the first thing searched. You can't search a subset of it. You can only make a subset out of the returned results.