> Hi Tatsuo Ishii : > > Thank you very much for your reply. Somehow I managed to restore the data by > creating by database in SQL_ASCII encoding. Then I changed the encoding of > my database into EUC_CN thru "update pg_database set encoding=2". But the > same problem happened in "vacuum verbose analyze". I went thru the changelog > and found out that you have add the multibyte validation feature from > release 7.2 on. Looks like the only way to get around is to patch the > source. > > But I would like to suggest that this feature should be made configurable. > You know encoding is a such a mess in the Chinese language, the popular > Chinese input methods actually allow the input of any Chinese character from > different encodings(e.g., GB2312/GBK, big5), it is difficult to make sure > that the user always input valid characters unless you validate the insert > and update clause.
But the encoding validation is made for such cases. It will prevent the database from INSERT/UPDATE with wrong encoding data. If you really want to allow input any GB2312/GBK, big5 etc. encoded data to the database, why don't you simply use SQL_ASCII or some other single byte encodings? If you mix up GB2312/GBK, big5 in a database you could not perform sort or any other database operation properly anyway. -- Tatsuo Ishii ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster
