Re: [cx-oracle-users] Antw: Bad conversion of a unicode value?

[matilda m. <ma...@gr...>]

>>> Michael Schlenker <msc=40contact.de> 27.11.2007 10:49 >>>
>
>SQL> select asciistr(unicp), unicp from nls_testing;
>
>ASCIISTR(UNICP)                                    UNICP
>-------------------------------------------------- --------
>=5C20AC                                              =E2=82=AC
>=5C2456                                              =C2=BF
>
>SQL> commit;
>Transaktion mit COMMIT abgeschlossen.
>
>As told, i get the first one correct, but the second does
>appear to not reach python intact.

Hi Michael,

just a short information in between. At the moment I can only
investigate
this problem in spare time. That=27s the reason that it will last some
time
anyway.

What I did: I created a short Embedded SQL C program to get the values
from the database. Interestingly I get the same byte sequences as I
get
them from python. My conclusion: There must be something with the
internal character set conversion between DB backend and DB client.

I also don=27t know at the moment how much the client environment (Linux
for me)
influences the results. E.g. I get =22rubbish=22 for both values stored in
your
test tables.

If you have new informations please let us know. Anthony is interested
in=20
improving the i18n and character set stuff in cx_Oracle. At the moment
he
seems a little bit busy. :-)


Best regards
Andreas Mock