Re: Per-column collation, proof of concept

2010/7/14 Peter Eisentraut <peter_e@gmx.net>:
> On ons, 2010-07-14 at 19:35 +0200, Pavel Stehule wrote:
>> I have only one question - If I understand well you can use collate
>> just for sort. What is your plan for range search operation?
>
> My patch does range searches.  Sorting uses the same operators, so both
> will be supported.  (Sorting is not yet implemented, as I had written.)
>
>> Sort is
>> interesting and I am sure important for multilangual applications, for
>> me - more important is case sensitive, case insensitive, accent
>> sensitive, insensitive filtering - do you have a plan for it?
>
> You may be able to do some of these by using appropriate locale
> definitions.  I'd need some examples to be able to tell for sure.
>
>> what about DISTINCT clause, maybe GROUP BY clause ?
>
> DISTINCT and GROUP BY work with equality, which is not affected by
> locales (at least under the current rules).
>

:( maybe we have to enhance a locales - or do some work in this way.
In Czech's IS is relative often operation some like

name = 'Stěhule' COLLATION cs_CZ_cs_ai -- compare case insensitive
accent insensitive

PostgreSQL is last db, that doesn't integreated support for it

Regards

Pavel

>
>