Stefan Probst wrote: > On the other side, there seems actually no need to display > non-NFKC for the Web, since as far as I understand, W3C > is planning to make NFKC a requirement for the Web.
Wait: my understanding is different here. A sequence like "q" + combining tilde *is* NFKC. It is also NFC, NFD and NFKD. What (canonical or compatibility) *composition* normalization does is converting a sequence to a precomposed character, *if* one exist. What W3C says when they mandate composition is that a sequence like "a" + combining tilde has to be converted to the single code point "a with tilde". But, if there is no precomposed character for "q with tilde", then the combining tilde *must* be maintained in all normalization forms. Whether or not this combining tilde is displayed nicely is a different story, but it cannot be discarded. > By trying to normalize the input (the combining sequences > to NFKC) IE might even work against planned W3C rules. Why? Isn't that what W3C asked? The only risk is that the normalization is a waste of time, because it has already been made by the server. BTW, are you sure that it is NFKC? My understanding is that it was NFC + some extra passages. For instance, would superscript numbers like "²" be turned to "2"? I hope not, as that would break lots of web pages. > Assuming, that the renderer is part of the OS and used by > most - if not all - applications, I conclude, that Windows > ME is not able to handle the combining modifier characters. The renderer is a DLL called Uniscribe, which is not shipped by default with all MS operating systems. I think that 95, 98 and ME only receive by installing IE support for some languages. > Anybody experiences with other OSs / other characters? To my experience, most Windows NT apps behave like your description: things are better only inside IE. But Windows 2000 seems to work fine out of the box, in most applications. _ Marco