Summary

Define a means for applications to specify property files encoded in UTF-8, and extend the ResourceBundle API to load them.

Motivation

The platform has for a long time provided a properties-file format that is based on ISO-8859-1 and provides an escape mechanism for characters that cannot be represented in that encoding. This format is supported by the standard resource-bundle lookup mechanism. As noted in the related RFEs, this format is difficult to use because it requires continuous conversion between its escaped form and text in character encodings that are directly editable.

Description

Change the default file encoding for ResourceBundle class to load properties files from ISO-8859-1 to UTF-8. By doing so, applications no longer need to convert the properties files using the escape mechanism. Existing properties files are rarely affected by this change, since ISO-8859-1's U+0000-U+007F are compatible with UTF-8, and characters whose code points are over U+00FF should have been escaped. If an exception occurs on reading a properties file in UTF-8, either a MalformedInputException or an UnmappableCharacterException, the properties file is read again from scratch, reverting to using ISO-8859-1 encoding. In order for a rare occasion where a ISO-8859-1 properties file can be recognized as a valid UTF-8 file, this JEP provides a means to explicitly designate the encoding either ISO-8859-1 or UTF-8, by setting the system property "java.util.PropertyResourceBundle.encoding".