We currently develop a kind of Portal, like a webtop...all the navigation is done through Ajax requests Initially, all the source files for the project are stored in a Subversion repository : the container has an UTF-8 character set. Initially our database, Oracle, was positionned to UTF 8 . Recently , our administrator moved the data to another Oracle instance in ISO 8859-15.
When playing the application, all the words with accents are not well rendered in our browser.
First, we put a pre filter for all incoming params (sent through a form_remote_tag), converting them from utf-8 to iso 8859 15 Our problem now is output. As Prototype only supports UTF 8, my idea was to convert all the source files to ISO (as the database) then create a post filter to convert all the content to UTF 8 (as we would do to gzip output).
Another idea could be to convert all the data to UTF8, coming from the database (iso), through find methods from our models, or custom SQL queries. Then , we could do the mix with partials, left in UTF 8 in the repository
What could be the best approach? (if you think like me 'leave the
database in UTF8, please don't reply 
we couldn't..this database is
a legacy database, and other older client server application based on
it need the ISO encoding )
Laurent