William,I agree with Tex that William's is an abhorrent suggestion.
You understand Unicode well enough by now, to know that this is an abhorent suggestion.
As the characters can be represented in Unicode by using Cyrillic plus combining diacriticals, to create a proprietary set of codes in the Private Use Area would introduce incompatibilities with other applications that support these characters in the recommended form. Following your recommendation would cause searching, sorting and interchange of Vladimir's data to fail in applications that properly support these characters. And it is likely difficult to get other applications to buy into supporting a proprietary solution. It is easier to address the rendering problem that Vladimir has than to unravel the mess your suggestion would create. It isn't even a good recommendation for short term use.
Did I miss something? Why are you recommending the PUA for this use?
tex
But one part of it might be rescued, if I understand rendering technologies correctly. Internally within a font only, Vladimir may define glyphs for his precomposed characters and assign PUA code points to them. In fact this seems to be how MS deals with some Hebrew presentation forms (ones which aren't in FB1D-FB4F) in Times New Roman. Then it can be part of the rendering technology Vladimir uses to substitute for the standard Unicode for his letter-accent combinations the PUA codes for the precomposed glyphs. For example, on a Microsoft system this may be included in an OpenType table, and Uniscribe (but only the soon to be released version which can process Cyrillic text) will perform the required glyph substitution. The PUA codes would never be used outside the font and rendering processor, and so their use is appropriately private. But then in such a system OpenType and Uniscribe could be set up to position the diacritic correctly without a precomposed glyph.
One further advantage of this over any approach using PUA for data interchange is that users who don't have the appropriate font and version of Uniscribe, or are using an OS which doesn't support this or an equivalent, will be able to read the text (they won't get spaces or square boxes) even though (as in Vladimir's examples) the diacritic placement will not be ideal.
-- Peter Kirk [EMAIL PROTECTED] http://web.onetel.net.uk/~peterkirk/
