Roland Kuhn wrote:
How do you know you're dealing with Unicode in the first place? Imagine a latin1 file name which incidentally does not violate the UTF-8 rules, but happens to be not normalized. Normalizing it will simply destroy it.
This is incorrect. NFC does not alter ISO-Latin-1 sequences. Any names constructed using ISO 2022 character sets will not contain C1 controls that are used by UTF-8 to indicate multibyte sequences. Where there would be a problem is with Microsoft ANSI code pages which put printable characters in the C1 range but on Windows we are going to have much broader problems due to the conversion from Code Pages to UNICODE which will make all previously written files inaccessible. _______________________________________________ OpenAFS-devel mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-devel
