- From: <juanrgonzaleza@canonicalscience.com>
- Date: Tue, 14 Mar 2006 11:09:52 -0800 (PST)
- To: <www-math@w3.org>

>> Bruce Miller said: > >To get to what seems to be your point: > [snip] > >I'm not sure what more I can say here; I'm not even sure >what the question is. Is it whether I prefer <mfrac> to <frac> >or <fraction> or even <f> ? Is it whether I prefer prefix >to infix (or more likely "natural" notations with mixtures of >infix, prefix, postfix, all with various precedence, sometimes >whose precedence depends on context...)? > It appears as you are attempting to trivialize the research. Do you really believe that important questions are if one prefer <mfrac> to <frac> or <fraction> or even <f>? >First, I would reinforce David's comments about mixed markup. >To take one of the examples from your posting: > a<frac/>{2 + b <frac/> c} >The DOM for that is rather strange, isn't it? > <math> > TEXT: a > <frac/> > TEXT: { 2 + b > <frac/> > c } > </math> Yes the DOM like strange. That is reason I said one could probably prefer grouping via explicit tag instead of using { and }. But the tree structure for example for <CanonMath>a<frac/><g>2 + b <frac/> c</g></CanonMath> is rather natural <CanonMath> a <frac/> <g> 2 + b <frac/> c </g> </CanonMath> from a *document* point of view. In fact, I explained how that kind of markup (namely mixed markup) is formally similar to usual HTML documents (look examples I introduced, the figures of trees and the comparison with a typical <i></i> HTML example). MathML is based in pure markup or “data oriented”. >Rather than pretending to use XML, wouldn't it be better to use >Unicode or some other character entities, instead of the empty elements? >You seem to be claiming that it is trivial to convert these forms to MathML >with XSLT; Have you actually tried this? Then empty elements are permitted in Content Math and XHTML but are not in others ways? I am not sure, but apparently you find a great disturbance reading some like <CanonMath> <textnode>a</textnode> <fraction/> <textnode>b</textnode> </CanonMath> but not trouble with <math> <apply> <divide/> <ci>a</ci> <ci>b</ci> </apply> </math> Yes, You are right, the abandon of XML and the use of textual islands as in LateX code embebbed into HTML or XML documents is an option. However, IteX, TeX/Latex, and ASCIIMath formats are not suitable input sintaxes for us. In my reply to David Carlishe I addes some examples of XSLT. What is more, I know people who has translated such as $$ 1 \over {x+1}^2 $$ to presentation MathML using XSLT. It would be not difficult to adapt the XSLT to the case <CanonMath>1<fraction/>{x+1}^2<CanonMath> or maybe <CanonMath>1<fraction/><g>x+1</g><power/>2<CanonMath> >If you move this form more towards XML, to clarify the grouping and such, you've >ended up with something surprisingly close to MathML; Where's the benefit? I think that one easily see next benefits with proposal: - Easier input syntax - single markup instead of double (presentation AND content) - possibility to extension to science that has not MathML was not designed for that (therein the developing of many scientific XML languages such as CML) - I already explained trouble I experienced with MathML tools (incorrect output, lack of semantics, bugs, needs of fine-tuning by hand, etc.) - Examples such as that of Hermes I detailed would be not permitted in new sintax is being designed. This favours a clean web, which is one of pillars of the vision of the w3c. Or you are supporting the use of TeX, which is next transformed into MathML + XHTML via Hermes for outputting empty paragraphs for layout and authors or dates encoded as headings of level 3 both deprecated and discouraged by last W3C specifications and recomendations? - Etc. Even in the adverse case <CanonMath><_>a + b<fract/>c + d</_></<CanonMath> (or similar) one still prefers over <math xmlns="http://www.w3.org/1998/Math/MathML" display=“block”><mfract><mrow><mi>a</mi><mo>+</mo><mi>b</mi></mrow><mrow><mi>c</mi><mo>+</mo><mi>d</mi></mi></mrow></mfract></math> or not? Note that there is not a great increasing on the complexity of input syntax of CanonMath when compared to pure TeX <_>a+b<fract/>c+d</_> \fract{a+b}{c+d} In other cases I listed, syntax is far more easy that with TeX/LaTeX. >OTOH, if you move this form towards a pure text, you've ended up >with something surprisingly close to Asciimath. Are the problems >with AsciiMath really unsolveable? (whatever they are; I'm not that familiar) > As already explained, ASCIIMath is a nice textual syntax with a well-defined niche. The modification of ASCIIMath for more general purposes would finalize with some completely different therefore I see not the point. If, however, if you are asking for the possibility for a textual version of CanonMath can be enclosed in a XML document Some as <CanonMath>{a + b fract c + d}</<CanonMath> it could be. I also am valuating that possibility. Then would I also abandon the XML scientific module by a textual version? And what about the CanonText module? Also abandoned and just transformed into XHTML on the fly? Hum, interesting suggestion... In fact, many people dislike XHTML angles and is working with alternative syntaxes. The Wiki is a popular example. >More generally: I don't believe it is fair to demand a response from >any particular person or group. I think the best you can do is to post >a clear and concise request for comments to some public forums (fora?) >[www-math is such a forum; it has not only Math WG members, but >many other parties interested in math, math on the web, etc, >who are free to comment, or not, as they wish.] >with the hope of finding people interested in your approach that are >willing to contribute. Alternatively, try to ask specific questions. > Anyone is free to read, comment, correct my errors, provide suggestions to main points I are highlighting. In fact people, is doing both in private or in public. > >-- >bruce.miller@nist.gov >http://math.nist.gov/~BMiller/ > > Juan R. Center for CANONICAL |SCIENCE)

Received on Tuesday, 14 March 2006 19:10:21 UTC