To be fair it only starts complaining about case-insensitivity and later really complains about insecurity through matching important filenames via ignorable code points. Maybe the problem is having multiple places that implement the rules for filename comparison.
It's true that doing an exact byte-for-byte match is easier, but I do like that I don't have to do over the phone support for parents without worrying about another file with the same name and differing case.
At least he acknowledges his design bugs of unicode identifiers. Strings are not garbage, even he likes them this way. Filenames need to stay identifiable, and thus need to adhere to unicode rules. TR 39. Case-insensitivity aside.
To be fair it only starts complaining about case-insensitivity and later really complains about insecurity through matching important filenames via ignorable code points. Maybe the problem is having multiple places that implement the rules for filename comparison.
It's true that doing an exact byte-for-byte match is easier, but I do like that I don't have to do over the phone support for parents without worrying about another file with the same name and differing case.
At least he acknowledges his design bugs of unicode identifiers. Strings are not garbage, even he likes them this way. Filenames need to stay identifiable, and thus need to adhere to unicode rules. TR 39. Case-insensitivity aside.
https://github.com/rurban/libu8ident