Martin Kochanski wrote: > And (3) it is not yet realistic to expect that > all software will be able to handle the non-1-to-1 relationship > between characters and glyphs, and it won't be for some time.
This argument belongs into the 1980s, not into the 2000s. It is one reason why Unicode has precomposed characters (other than the argument of easy round-trip conversion, etc.) and needs normalization and such. > You need an application that understands the distinction; Almost all applications do _not_ handle text on such a detailed level. If they do anything more interesting than shuffling strings from resource bundles or user input to the display APIs, then applications use libraries to handle complicated stuff. > an operating system that supports it properly; MacOS X and Windows 2k/XP are quite good at it. Linux is getting it. > and display and printer drivers that don't corrupt the result. What the OS does not do you can do with off-the-shelf layout engines like the one in ICU, which is open-source. http://oss.software.ibm.com/icu/userguide/layoutEngine.html Such libraries for layout, like for any other I18N/Unicode work, avoid everyone reinventing the wheel. Best regards, markus
