Dmitry Vereschaka <[EMAIL PROTECTED]> writes: > HN> A program can be built on one machine and run on one or more > HN> others. Regardless, I'm not aware of a way to reliably > HN> determine the coding of a file system. (Other than the simple > HN> Windows vs. Unix, although even that fails in presence of SMB > HN> shares, Cygwin, and the like.) > > Why not to invent new switch --force-filesystem-encoding ? :)
Or a charset extension to --restrict-file-names. If someone contributes such a change, it would be worthwhile to consider.
