Peter Eisentraut <peter_e@gmx.net> writes:
> Am Montag, 5. September 2005 17:40 schrieb Harald Fuchs:
>> I dunno the details of the supported encodings, but is there
>> any encoding where 'a-z' includes more or less than 26 letters?
> Well, it seems that our regexp library interprets [a-z] as exactly 26 letters,
> but that seems to be a lack of locale support rather than a feature.
ISTM that the notation should properly mean "every character that sorts
between a and z inclusive". So it's sort order dependent and thus
locale dependent --- or at least should be. The locale support in our
regexp code is definitely pretty weak at the moment.
regards, tom lane
