Re: Can someone please explain this misbehavior?

[A B]

"Ё" is not "accented", it is a separate letter in Cyrillic alphabet. Same for "й", which isn't threated as "accented" by MySQL for some reason.

SET @l1 = CONVERT(CAST(0xd0b8 AS CHAR) USING 'utf8') COLLATE 'utf8_general_ci';
SET @l2 = CONVERT(CAST(0xd0b9 AS CHAR) USING 'utf8') COLLATE 'utf8_general_ci';
SELECT @l1 AS `l1`, @l2 AS `l2`,
IF(@l1 = @l2, 'Bogus comparison', 'Can\'t be more correct') AS `equality`;

And some other letters.

Strictly speaking, the first word doesn't even exist. (Guess, why it's a bogus comparison?)

It's difficult, but possible to imagine, why _general collation threat [seemingly] accented letters as equal, but I would expect utf8_unicode_ci to be more sensitive to the details.

This MySQL behavior ruining unique keys when converting tables from cp1251(and a number of other cyrillic encodings) to UTF8.