Originally posted by tny
Case sensitivity is bad.
On a filesystem that supports UNICODE, case insensistivity is almost impossible.
First of all, on HFS+ right now I can create file names that look *identical* using differing character sets (i.e. mixing greek, cyrillic, and roman character sets) in a file name.
Secondly, it imposes unrealistic limitations for languages or character sets where case can have huge effects on meaning. For instance, some languages may use an upper or lower case characters to denote a part of speach, such as distinguishing singular from plural, subject from object, nouns from verbs, or 1st person from 3rd person. For instance, in some language, it may be impossible to have singular and pluralized filenames in the same directory.
Thirdly, European languages are discriminated against because there are no limitations placed on filenames that use Asian or Middle Eastern character sets.
In an idealized world, a file system should be a storage place for binary data, the filename and metadata should be irrelevant and be abstracted at a higher level, like in a relational database. Then maybe we will see localizable filenames (i.e. one file can show different localized names depending on the preferences of the user) and rich extensible metadata which we don't currently see with today's filesystems.
