> Looks like there is no resolution but replacement. Thanks. Yes, some overly generalised statements follow. A worthy endeavour is to educate your users or use tools that prevent the problem.
Unless the entire software stack from the lowest level is encoding aware and understands and carries correctly the complete information in your preferred encoding end to end, you're inevitably going to have problems at some stage or abstraction layer. The reliable solution is to use the common denominator in carrying names (any information) between software implementation. As majority of the software is alpha-numeric English symbols aware (only) including the language used for programming, logically either convert it all to your encoding (properly end to end, portable) or use a naming scheme that is a subset of your encoding to match the common denominator. In simple words: use 7-bit old school ASCII. There are packages to help you convert, or simply start over. Another attempt at this is UTF-8, which may not solve it for you or create a new set of problems. No matter nfs, cifs or whatever (future (proof (really))) protocol you use, this problem will always be present and simply resolved by not creating the use cases outside generic symbol usage.
