- everything is in ascii unicode escape sequences all the time (is this what we have now?)
yes, this is what we have. use \uXXXX ascii as the i18n resource storage format.
- everything is in utf-8 all the time (svn, checked out source code, any compiled property files
not exactly, we can't use utf-8(except 8859-1 char) in compiled properties because resource bundle only happy with \uXXXX escaped format with 8859-1 charset.
- utf-8 in svn and checkout and unicode escape after compilation
This is what I'm going to do.EOL of svn does not have any relationship with this i18n content encoding. I've tested Eclipse, notepad, gedit, vim. They are all happy with UTF-8 i18n properties. I think if the editor support utf-8, there's no reason it can't read UTF-8 i18n native language properties.
As for the question "why would we change to ascii during compilation?"
The problem is with resource bundles. For resouce bundle, Properties files are always read as ISO-8859-1. As a result, To include Unicode characters in i18n properties file, we must use \uXXXX escapes(with native2ascii tool).