A
Aidan Jeffery
Guest
ICU - International Components for Unicode. Try: http://site.icu-project.org The default collation for the utf/empty databases is Basic, which resorts to binary sorting of characters above the ASCII range. This is not ideal for queries where non-ASCII characters are involved, because they have to be resolved on the client. A set of ICU collations is provided with OpenEdge under DLC/prolang/utf. If you don't have a sorting requirement for a particular locale, ICU-UCA is the best choice. Use Data Admin to load the collation definition (e.g. ICU-UCA.df). You have to rebuild the database indexes immediately after doing this. > I think there has been problems in the past where certain characters are not properly converted to UTF-8. This would be a concern. Is it possible that the database(s) contain characters that are not correctly encoded in iso8859-1? If the system has Windows clients, could some 1252 characters have got in there? It would be good to run a proutil convchar charscan, then deal with any problems that are reported. As for the speed of loading the data, have you created adequately sized fixed extents for the larger tables before loading?
Continue reading...
Continue reading...