No, it makes much more sense to default to the "natural" charset of 
the user's language, because this is the encoding that most people use 
in their data. Exporting and importing from Turba is actually not a 
very common task.

According to Jan Schneider, addressbooks (except in vCard format) are 
exported in the encoding that is used for the user interface (UI), i. 
e. UTF-8 in most cases. So, most addressbooks available for import 
will be encoded in UTF-8 (except, of course, pre-H3 address books).



Now, in the "Select the charset" field in the import form, the 
$my_charset entry is preselected, where $my_charset is the original 
charset of the translation (as set in turba/data.php). This 
preselection may cause user errors, in the situations outlined above. 
Note that apparently the $my_charset variable is used nowhere else.



It is suggested that the "Select the charset" field should rather have 
the UI encoding preselected, so addressbooks exported from Turba H3 
will be correctly imported with the default setting.



Attached is a patch to implement this suggestion. As $my_charset is 
only used in this single instance, I propose to mend its definition in 
turba/data.php. If you would rather mend the selection widget in 
turba/templates/data/import.inc, please ask for the pertinent patch.