Re: Extensions, patch v20 (bitrot fixes)

On Tue, Dec 21, 2010 at 08:04, Martijn van Oosterhout <kleptog(at)svana(dot)org> wrote:
> On Mon, Dec 20, 2010 at 10:15:56PM +0100, Nicolas Barbier wrote:
>> >From <URL:http://en.wikipedia.org/wiki/Japanese_language_and_computers#Character_encodings>:
> ISTM that since all the mapping tables are public it should be a SMOP
> to *prove* roundtrip conversions are safe, or identify the problems.

Another issue in Japanese users is EUDC (End User Defined Character).
Unfortunately for both postgres developers and application developers
in Japan, many machine dependence characters are still used in popular
mobile phones in Japan. Their native encoding is SHIFT_JIS, and we
have an EUDC mapping for SHIFT_JIS to/from EUC_JP. But we don't have
for UTF-8 to/from other encodings. That is one of the reasons why we
cannot move to the UTF-8 world completely.

Imagine that a module that manipulate EUDC text. It will be written
in EUC_JP because SHIFT_JIS is not supported in postgres. Also, it
cannot be rewritten in UTF-8 because there are no mapping for the
characters used in it.

--
Itagaki Takahiro