At my ISP I have a 4.0.26 MySQL server running Latin1 as default charset. If I specify ASCII or DEFAULt it works with my nationals æøåÆØÅ. If I choose LATIN1 then some other glyphs appear! This is the same for HTTP-connection and direct connection.
Also the same on MySQL 5.0.19 on my local
I think I can recognize a czech 'r-with a-hacek 'and that character/glyph belongs to LATIN2 (that is not available) I believe. When I enter a Danisk 'ø' is is immeditately transformed to a 'r-with-a-hacek' when saved.
HISTORY on 4.0.26 shows:
QUOTE
Set character_set_results=latin1
HISTORY on local 5.0.19 shows:
QUOTE
Set character_set_connection=utf8
Set character_set_client=utf8
Set character_set_results=latin1
Set character_set_client=utf8
Set character_set_results=latin1
Now as this server is running uft8 right now this is what one would expect.
After changing server default to latin1 HISTORY shows
QUOTE
Set character_set_connection=latin1
Set character_set_client=latin1
Set character_set_results=latin1
Set character_set_client=latin1
Set character_set_results=latin1
and the display is the same. No 'ø' but a 'r-with-a-hacek'.
However choosing 'DEFAULT' from connection manager HISTORY shows:
QUOTE
Set character_set_connection=latin1
Set character_set_client=latin1
Set character_set_results=default
Set character_set_client=latin1
Set character_set_results=default
and 'ø' is displayed
'ø' is only chosen as an example. 'æ' and 'å' also displays as some central/eastern European characters (a Polish accented l (pronounced as 'w') for instance!
If I manully issue 'Set character_set_results=latin2' those characters are displayed as '?' and in a HEX-editor they ae identified as char(63) after SQL- and CSV-export.
To the extent that I can verify other charset settings are OK (cyrillic cp1251, Arabic tis620)
So I really do not understand!