We would rather not use anything that tries to guess encodings. There is no reliable way to do so and it can go horribly wrong and cause more problems that it solves.
Explicit encodings work best, either though configuration (as you can do with the Text File Thumbnails plugin, but not currently with the Text Viewer plugin; we may merge the two at some point), or by using Unicode with a BOM at the start of the file, if the file is not in the codepage that the OS is configured to use.