Subject RE: [firebird-support] Filter non ISO8859_1 characters
Author Svein Erling Tysvær
> I have a FireBird 2.5 database, created to use the ISO8859_1 character set. On a monthly interval,
> I re-create a new empty database (except for a few lookup tables) by restoring a backup, dump data
> from an Oracle database and load the mentioned empty database. I then do some indexing using Href
> Tools Corp. Rubicon indexing components (the same one used on My web
> application is
> It now seems that non ISO8859_1 characters are slipping thru and they are causing problems during
> The Rubicon indexing.
> Does anyone know how can I filter out any non ISO8859_1 characters during the load process?
> I'm using Delphi7 and the UniDac database drivers.

Hi, Nols, since no-one has answered, I'll try (even though I don't know much about the subject and nothing about UniDac).

Have you specified ISO8859_1 both when creating the database and when connecting to it for the import? I must admit that I am sloppy when writing IBO applications and only occasionally set the CharSet property on my TIB_Connections. And what about collation? I don't think there should be any characters that are available to one collation and not to another, but it certainly influence how things are ordered (some collations are case sensitive, other insensitive and as a Norwegian, I order my alphabet slightly different from what Swedes would do).

Hope this either helps you or makes someone else aware of this thread and gives a better answer,