[TYPO3-core] The installer, the charset and the DB

[Dmitry Dulepov]

Hi!

Martin Kutschker wrote:
> So why don't we switch to PHP 5.1 right now? Seriously, we have a plan 
> to do a migration to newer software for 5.0 and I really think it sucks 
> when I have use UTF-8 just because some TYPO3 devs think it is so cool.

I never saw (may be I missed) descriptions like "this is a PHP 4.x bug" 
in the bug tracker but I saw several like this for PHP5. This prevents 
*me* from switching to PHP5. I feel that it is not as stable as PHP4. 
May be I am wrong.

> I do a dump and then use "recode" using windows-1252 as source charset. 
> You can use iconv for sanity checks, cecause it will choke on illegal 
> chars (that's why I use windows-1252!).

I had 3.23 + utf8 content over latin1 database. Characters were latvian: 
ēūīāšģķļčņ. If I remember correctly, I could get ļ and one another 
character imported correctly. I always got question mark. Other 
characters went ok.

I tried converting from utf-8 to windows-1257 (native latvian) with 
iconv to see if it works (just to check that sql export was ok. It was 
ok. I tried to convert back and got identical utf-8. I tried to import 
it again using two different ways (see below) but still could not get 
characters right.

The ways I used:
1. mysql -u xxx -p -charset utf8 dbname
2. set names utf8; set session character_set_server=utf8

I understand that I might've done something incorrectly. But someone 
else could go my [wrong] way too. So detailed manual would be really 
excellent.

Dmitry.
-- 
"It is our choices, that show what we truly are,
far more than our abilities." (A.P.W.B.D.)