Hi Hiroshi,
I did some test with Win ME and the same application I used for the
previous tests with XP. Just note that my application is quite old and
was developed for Win NT and Win 98. Before the psqlodbc+ driver came on
the scene I was forced to use the ANSI-Version of the driver 8.01.XX.
Set-up is the same as before (ME instead of XP -- Debian Sarge -- PG
7.4.X). I tested with LATIN9 and UNICODE database, verified with psql
and pgAdmin (using XP of course).
Again I tested the complete ISO-8859-15/Latin9 character set ( 0x20 to
0xFF without unprintable char.). It worked, excluding the bytes 0xA6,
0xA8, 0xB4, 0B8, 0xBC, 0xBD, 0xBE. Obviously my application was written
for ISO-8859-1/Latin1 with just the Euro-character added (0xA4). The
same occurred on Win XP. But I didn't noticed during the first tests. It
seems, my application just can't read these bytes, since checking the
same data through pgAdmin doesn't show any errors.
Please also not that I just used an updated version of Win ME and but
didn't update the driver manager. I also didn't use "set client_encoding
to XXX". But the driver functioned smoothly with LATIN9 and UNICODE
database. Also, no crashes occurred.
IMHO we don't need two different driver versions any more. Good job,
Hiroshi (Inoue and Saito)!
Any comments? Questions? Any further tests required?
Regards,
Johann