Hi, Peter, Dave!

Unicode standard defines default case mappings: see UnicodeData.txt and
SpecialCasing.txt
I don't believe any sane person expects anything different from default
case conversion.

Existing firebird behavior to uppercase ASCII7 characters only looks and
smells like bug in nowadays world, I'm sorry.
This is why preserving backward compatibility with this issue IMO
doesn't worth the effort.

At the same time having full non-binary default collation for sorting
purposes doesn't look feasible.
Even for simplistic WIN1251 charset there is a handful cultures using it
and all have different codepoint ordering rules. Defaulting to largest
culture looks like violation of political correctness. :-)

My opinion is that we should provide case conversion routines for
default binary collations based on Unicode mappings and do not worry
about backward compatibility here. Since UPPER/LOWER stuff is not stored
in indexes we are not breaking ODS consistency and we only give users
possibility to use case conversion in cases where it didn't work before.

Nickolay Samofatov