Re: UTF-8 and LIKE vs =

On Aug 24, 2004, at 7:21 PM, Tatsuo Ishii wrote:

> I don't know exactly what kind of encodings you wish to use, but I
> think MULE_INTERNAL might help you. It's actually mixture of various
> encodings with encoding-prefix added to each letter.  For example if
> you can mix KS5601 (Korean) and Japanese Kanji (JIS X0208) in a *same*
> column. So if you don't have problem with sorting KS5601/JIS X0208
> with C locale, you should not have problem with MULE_INTERNAL too in
> theory.

MULE_INTERNAL? Is that a locale?

> Remaining problem is how to display the Korean-Japanese mixed string
> in your client, but this is not PostgreSQL's problem, of
> course. However you could write your own conversion function
> MULE_INTERNAL <--> UTF-8, and might be able to solve the problem.

Well, the strings aren't usually mixed, but there can be different
languages in different rows. For example, I have a customer with a
single Bricolage instance in which they manage content in all of the
following languages:

   Burmese
   Cantonese
   English
   Korean
   Khmer
   Lao
   Mandarin
   Tibetan
   Uyghur
   Vietnamese

And they might do a search that returns results in more than one of
these languages.

Regards,

David