Paul Ruizendaal wrote:

>> The other is to reduce the size of
>> generated keys when the range of code points is known. 8859-1, for
>> example, can't be represented in single byte utf8, but base characters
>> can be represented in a single byte.
>
> Sounds like "bit bumming" to me. Why is this optimisation relevant in
> today's world?

It is precisely bit bumming, but the size of index keys matters a lot
more than the stored size of data. Multi-level collations add a byte
per character per level, so a relatively modest 250 character field
can produce a key that 1000 bytes long or more.


Cheers,

Ann