Sorry, I guess not all mailers handle embedded graphics in HTML messages. I posted it so that you could see the graphics, on
http://www.macchiato.com/utc/variation_selection/variation_selection_followu p.htm It is *not* exactly the same. I added and rearranged the concrete examples at the very end. I think those last examples of the font are key to this issue: depending on which we say conformantly support the variation sequence will help determine how we handle this issue. I also posted my previous paper, although my thinking has changed a bit since then, particularly on the 'tightness' of the description -- the paragraph containing "Suppose a glyph has a slash with an angle of 32° from vertical." http://www.macchiato.com/utc/variation_selection/variation_selection.htm Mark ————— Πόλλ’ ἠπίστατο ἔργα, κακῶς δ’ ἠπίστατο πάντα — Ὁμήρου Μαργίτῃ [For transliteration, see http://oss.software.ibm.com/cgi-bin/icu/tr] http://www.macchiato.com ----- Original Message ----- From: "Asmus Freytag" <[EMAIL PROTECTED]> To: "Mark Davis (jtcsv)" <[EMAIL PROTECTED]> Sent: Sunday, January 27, 2002 01:36 Subject: Re: Variation Selection (Was Re: Unicode 3.2: BETA files updated) Now this message was clear as mud. Literally. As in totally opaque in the pure visual sense: All the examples show as nice gray boxes in my mailer - please make sure whatever you are sending is an attachment, and not in-line. Thanks, A./ (PS: They may well arrive in viewable condition when this reply gets back to you..., but I can't read it) ————— Πόλλ’ ἠπίστατο ἔργα, κακῶς δ’ ἠπίστατο πάντα — Ὁμήρου Μαργίτῃ [For transliteration, see http://oss.software.ibm.com/cgi-bin/icu/tr] http://www.macchiato.com ----- Original Message ----- From: "Asmus Freytag" <[EMAIL PROTECTED]> To: "Mark Davis (jtcsv)" <[EMAIL PROTECTED]> Sent: Sunday, January 27, 2002 01:36 Subject: Re: Variation Selection (Was Re: Unicode 3.2: BETA files updated) Now this message was clear as mud. Literally. As in totally opaque in the pure visual sense: All the examples show as nice gray boxes in my mailer - please make sure whatever you are sending is an attachment, and not in-line. Thanks, A./ (PS: They may well arrive in viewable condition when this reply gets back to you..., but I can't read it) At 12:33 AM 1/27/02 -0800, you wrote: > > > > >The other possibility is to say that to be strictly Unicode-conformant, > > >fonts should always use the reference glyph for unmarked characters > > >(ignoring differences only of style). I think this is actually a better > > > > Boy, great minds to think alike. Mark Davis just proposed that in > > a paper to the UTC this week. > >I would like to thank you for the compliment, but I must not have been >clear in my paper, since that is not really what I was proposing. Let me >try again, with a concrete case. > >Since I don't have a variety of math fonts, I will use the example of >U+03B8 greek small letter theta. This character can be represented by the >following glyphs (collected on my system) and many more. Any of these are >acceptable, conformant representations of the theta, depending on the >design of the font. > >1f908bf2.jpg > >Now, there is already a presentation form for the "open" form of the >theta, at U+03D1 greek theta Symbol. A presentation form of a character >represents the same abstract character, but is restricted in format to a >subset of the possible glyphs that could represent the character, such as >the following: > >1f908db5.jpg > >But suppose there were no such character. In that case, we might add an >encoded variant, expressed as an entry in the StandardizedVariants.html >file in the UCD, looking something like: >Ref Glyph Character Sequence Alt Glyph Description of variant appearance >1f908e05.jpg 03B8, FE00 1f908e4b.jpg Open Theta, unconnected on the left side > >Now there are four key facts that we need to keep in mind: > * This does not, at all, prevent the character alone <U+03B8> from > having the same appearance as what is titled as the Alt Glyph. It is > still a perfectly acceptable representation of that character in normal text. > * What is titled Alt Glyph in this entry is also merely a > representative, one of many possible glyphs that can represent the > variant. Thus both of the glyphs are "representative": they do not, and > cannot, encompass all of the possible glyphs that can represent either > the character alone, or the variant. (It would be a good idea for us to > change the title of this column, since it may mislead some into thinking > that that precise glyph must be used: it should be Variant Ref Glyph.) > * A key feature of the entry is the description; it provides the > limitation on the set of glyphs that can be used to represent the > sequence, if the sequence is supported by a font. > * If a font does not support the variation sequence given by this > entry, that it is also perfectly acceptable to ignore the U+FE00, and > thus render the sequence <U+03B8, U+FE00> with precisely the same glyph > as the single character <U+03B8>. >So the open issue is: how do we specify variation selection in the Unicode >Standard such that people can make a distinction between <U+03B8, U+FE00> >and <U+03B8>? > >I could see two possible alternatives. >>(a) Add an explicit entry to StandardizedVariants.html, looking like the >>following: >>Ref Glyph Character Sequence Alt Glyph Description of variant appearance >>1f908e69.jpg 03B8, FE01 1f908ea5.jpg Closed Theta, connected on the left side >>(Where I differed from Asmus is he felt that we could reserve a VS for >>this purpose, without supplying an explicit entry describing the >>limitation on the possible glyphs. One could conceivably take it to be >>the negation of each of the descriptions of variants of that character.) >> >>(b) Make it clear that a font supports a variation selection sequence if >>and only if it provides a visual distinction between the character alone >>vs. the character followed by a variation selector; and that the visual >>distinction is based on the description of the variant appearance in the entry. >> >>This also goes for multiple sequences: if a font supports <X, VS1> and >><X, VS2>, then it must visually distinguish all three sequences: <X>, <X, >>VS1>, and <X, VS2>; and the distinction must be in accordance with the >>given descriptions. >> >>While I think that either of these two approaches are acceptable, I find >>myself currently leaning towards the second. I find it fairly pointless >>to say that a font "supports" the variation selection sequence <U+03B8, >>U+FE00> if it does not provide a visual distinction from <U+03B8>; and >>such a distinction should be based on the entry description. Thus, of the >>following four fonts, only number 4 correctly supports the sequence >><U+03B8, U+FE00>. (Of course, any real font would have designs for the >>two glyphs that were a bit more harmonious!) >>Font <U+03B8> <U+03B8, U+FE00> >>1 1f908ec3.jpg 1f908ee1.jpg >>2 1f908ef5.jpg 1f908f31.jpg >>3 1f908f6d.jpg 1f908f8b.jpg >>4 1f908faa.jpg 1f908ff0.jpg >> >>Mark >> >>â?"â?"â?"â?"â?" >> >>Î ÏOλλâ?T á¼ Ï?ίÏfÏ"αÏ"ο á¼"Ïγα, >κακῶÏ, δâ?T á¼ >>Ï?ίÏfÏ"αÏ"ο Ï?άνÏ"α â?" á½?μήÏοÏ. >ÎoαÏγίÏ"á¿f >>[For transliteration, see >><http://oss.software.ibm.com/cgi-bin/icu/tr>http://oss.software.ibm.com/cg >>i-bin/icu/tr] >> >>http://www.macchiato.com >>----- Original Message ----- >>From: "Asmus Freytag" <<mailto:[EMAIL PROTECTED]>[EMAIL PROTECTED]> >>To: "David Hopwood" >><<mailto:[EMAIL PROTECTED]>[EMAIL PROTECTED]>; >><<mailto:[EMAIL PROTECTED]>[EMAIL PROTECTED]> >>Sent: Friday, January 25, 2002 22:37 >>Subject: Re: Unicode 3.2: BETA files updated >> >> > At 10:58 PM 1/24/02 +0000, David Hopwood wrote: >> > >One possibility is to make VS1 specify what is now the reference glyph, >> > >and VS2 specify the alternate glyph. Unmarked would mean either. >> > >> > Boy, great minds do think alike. I proposed that in a paper to the UTC >> > last year. ;-) >> > >> > You realize that this issue is not limited to variation selectors? >> > Read the section on greek phi in >> <http://www.unicode.org/unicode/reports/tr28>http://www.unicode.org/unico >> de/reports/tr28 >> > >> > >The other possibility is to say that to be strictly Unicode-conformant, >> > >fonts should always use the reference glyph for unmarked characters >> > >(ignoring differences only of style). I think this is actually a better >> > >solution in practice; it avoids having to add selectors that would >> > >usually be redundant, and that would interfere with normalisation. >> > >It's also consistent with the Mongolian variant selectors, where >> > >unmarked should mean the "first form". >> > >> > Boy, great minds to think alike. Mark Davis just proposed that in >> > a paper to the UTC this week. >> > >> > Unfortunately. this is not a model that's always usable. Please >> > read the section on phi for background. >> > >> > By adding a variation, we cannot restrict the glyph range for the >> > unmarked character - Mongolian being an exception since the unmarked >> > character's glyph range has been *explicitly* restricted from the >> > outset to the standard positional forms. >> > >> > For VS1, the situation is different in that the glyph range of the >> > *unmarked* character *also* includes the glyph identified by VS1. >> > >> > A./ >> > >> >