Re: tsearch2 in postgresql 8.3.1 - invalid byte sequence for encoding "UTF8": 0xc3

patrick wrote:
>> Can you identify which row(s) are causing this problem? If we have the
>> value that's causing this, someone can reproduce it.
>
> i have only 1 row:
> 46; "the product name"; "the description";
>
> i don't see any specials chars or accents.

I think I've reproduced it here, and it's not your data.

> knowing that some of my clients are french, should i use LATIN9 as
> database encoding / client encoding? or maybe it's because of the LOCALE
> (French/Canada)?

> UPDATE product SET search_vector = setweight(to_tsvector(name), 'A') ||
> to_tsvector(description);

It's your script. It's got a character that's not UTF8 and you've told
the system that's what your client_encoding is. I think it's "|".

See if you can reproduce it with: SELECT 'abc'::text || 'def::text;

I got the error by editing the script at a command-prompt. If you use
notepad or some other MS-Windows (TM) based editor it should let you
choose ANSI as the format to save in.

--
Richard Huxton
Archonet Ltd