Oleg Broytmann writes:
On Thu, Sep 13, 2001 at 11:38:42AM +0200, Tino Wildenhain wrote:
Certainly it is possible. Oracle, for example, allows to create multilingual multiencoding databases. But it is not so simple.
Maybe its possible to define the splitter per field?
The problem is not in the Splitter itself - the problem is in locale data. Locale data is the accumulated result of long research. Many people collected the data (how to represent national symbols, how to sort them, which are printable, which are alphas, etc), organized it and put into locales. If you are going to get rid of locale - where you will get this information? You need it for every language/encoding you are going to support. Some time ago, I read the GNU locale implementation. They complained about the global-ness of Posix Locale support and I think they provided a more object oriented implementation.
Maybe, such an implementation could be used to switch Locales thread specific and thereby switch Locales as necessary for the application at hand. However, one would probably need a GNU C-library to make use of it. Dieter