I agree that elaborate PUA schemes will not see any public acceptance. The most practical use we've seen is as a holding area for characters transcoded from Asian character sets that are not in the Unicode Standard (they *may* end up in it in the future, or not; or may be represented as variant sequences (see U3.2).)
That mechanism does allow for round-tripping (at least internally, and *if* the text is not mixed with PUA characters from other sets). Not an ideal solution, but one that meets certain needs. See http://www.unicode.org/unicode/reports/tr22/#Completeness Mark ————— Γνῶθι σαυτόν — Θαλῆς [For transliteration, see http://oss.software.ibm.com/cgi-bin/icu/tr] http://www.macchiato.com ----- Original Message ----- From: "Doug Ewell" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Cc: "Michael Everson" <[EMAIL PROTECTED]>; "William Overington" <[EMAIL PROTECTED]> Sent: Friday, April 26, 2002 08:50 Subject: Re: Towards a classification system for uses of the Private Use Area > Michael Everson <[EMAIL PROTECTED]> wrote: > > > The Private Use Area is not to be classified. Anyone anywhere can use > > any of its code points for anything. > > Furthermore, even if William's scheme is intended to be a semi-private > "convention" rather than an official part of Unicode -- much like > Michael's (and John Cowan's) own ConScript Unicode Registry -- it seems > unlikely that an elaborate indexing scheme such as the one William > proposes would gain much of a following. Vendors have only recently > started to implement surrogates properly, and still balk at decoding > SCSU (which is easy; it's the encoding part that gets complex). And > these official Unicode mechanisms are simple compared to William's "hex > point" indexing scheme. > > Additionally, it is VERY important to repeat -- probably more important > than anything else in this discussion -- that there is no automatic path > to "promotion" of any private-use character to full Unicode status. > Every character and script that is encoded in Unicode must undergo the > same scrutiny, regardless of how (or whether) it may have been encoded > in the past. That goes for Deseret and Shavian (accepted) as well as > Klingon and Aiha (not accepted), all of which were encoded in ConScript > but none of which were automatically "promoted" on that basis alone. > > There are two full Private Use planes, 131,068 code points in all (not > counting the four noncharacters), certainly enough for any private-use > implementation that would be envisioned as benefiting from William's > proposal, and a lot easier to implement (and thus more likely to be > used). My suggestion to William is that if he envisions a potentially > widespread use for the PUA, he may consider creating a ConScript-like > registry for the upper planes. That would be just as effective and much > simpler. > > Apart from font vendors who use the PUA for presentation forms within > the font, what current practices exist for using the PUA? I mentioned > ConScript; how popular is its use? Are there any other commonly used > practices or conventions? Apple has blocked out a code point for APPLE > SIGN, and somebody (sorry, don't remember the name) mentioned a > Microsoft convention of a subarea for symbols or dingbats. Maybe a > discussion along these lines can reveal the true nature of PUA use and > help William redirect his considerable energy toward a more practical > system. > > -Doug Ewell > Fullerton, California > > > >
