* Lars Marius Garshol | | Well, that's what I thought, anyway. But now my Japanese colleagues | tell me that Japanese users sometimes have to define their own | characters in their own fonts in order to be able to write some | characters they may need (for example to write their own name), and | that the UDCs may be intended for this.
* David Hopwood | | Using an image (with alt tag giving a transliteration, say) is much | better, since it doesn't require anyone to have custom fonts. | | However, if anyone wanted to use the PUA for this, NCRs (, | etc.) would be a better way of doing it than relying on any | particular mapping from Shift_JIS to the PUA. I agree completely on both counts, but actual practice rarely seems to be informed by such considerations. My problem is that if people routinely do this I need to support it. (In the same way I need to interpret NCR references in the 128-159 range to refer to Win-1252 code points. Too many pages break if we do otherwise.) | My advice would be not to bother with supporting PUA mappings for | WWW use, and use the Unicode-provided CP932 mapping table for the | MIME charset "Shift_JIS". I'd be happy to do so, and unless people tell me that UDCs are actually used in practice by Japanese users, I will. --Lars M.