Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On Wed, 15 Feb 2006 14:26:39 -0300, Clovis Garcia <clovis at phoebus.com.br> wrote: >Bo, > >The INSERT command is already using de designator N before the column >Message, but the problem persists. I think the text is been corrupted >before the call of the command because the Trace shows the text already >changed. > How did you verify this? Did you look in the CVSNT sources or otherwise? I'd like to find where I could make a change and see what is happening, for example Tony says: <quote> it'll just take the unicode string it's passed and covert it to the database ANSI codepage first </quote> I think that this is actually the culprit, one should not modify/convert these strings at all. Just sending them to the database with the N prefix and into tables which have columns with nvarchar or ntext should suffice. But as I said I don't particularly know where the call to the ANSI conversion is done so I cannot find where to comment out the conversion and test. The interesting thing is: 1) The data supplied to the loginfo script is OK, all Swedish characters come through just fine both in the folder/file name and in the log message. 2) The RCS file likewise contain the wanted text in the log messages, no problems here either. So it looks like there is an unneeded/unwanted conversion prior to storing data into the Audit database that corrupts the strings. As a check I also fired up my Win2003 Enterprise Server where I have Auditing enabled via SQLite instead of MSSQL. This also has the same problems with corrupted file names and log messages when they contain Swedish characters... /Bo (Bo Berglund, developer in Sweden)