Looking at the code and tests, it seems to me that the primary difference in like processing for TERRITORY_BASED vs USC_BASIC collation is that processing is handled on collation elements rather than characters, so that _ will match a single collation element and not a single character. It seems that that particular behavior is a bug, _ should match a single character regardless of collation elements. So, my question is, how is like processing supposed to be different for TERRITORY_BASED vs USC_BASIC and can anyone give an example where the results should be different?

Thanks

Kathey


Reply via email to