From: Jens Jorgen Mortensen <jensj@fy...>  20041213 11:35:12

David Goodger wrote: > I have interspersed specific comments below. My general comment is > that I don"t think reST should invent a new math syntax. There are so > many details and intricacies, it"s opening a huge can of worms. > Instead, we should adopt one or more existing math syntaxes. I"d > rather see an exploration and discussion of the available, existing > syntaxes, available software, and output support options. > > There"s a practical reason for my position as well. Up to now, nobody > has had the time or will to implement a proper "itex" or "tex" or > "mathml" directive (except for Beni"s math hacks, but they"re hacks > ;). I very much doubt that anyone will jump on this opportunity to > do a lot more work for fewer rewards. > > You"re welcome to pursue this, of course. If you do, I hope you > succeed. But I have neither the time nor the desire to do much for > it. The discussion sort of stopped here! Probably because nobody wants to do the work. I am willing to do the work if we can agree on what to do. I already implemented a parser for a syntax similar to that proposed by Alan (http://sourceforge.net/mailarchive/message.php?msg_id=10020457) some time ago. The big question is: *Should* we have a reSTnative math syntax? If yes, what should it look like? I like the proposal of Alan  I am a little in doubt concerning the use of backtics for grouping. Grouping can be used for distinguishing: a_`i+1` and a_i + 1 Alternatives: * Use lack of whitespace as grouping: a_i+1 and a_i + 1 * Use "{}" for grouping (::math:`a_{i+1}`)  we loose "{" and "}" as normal characters. * Use "{}" for grouping only for subscript, superscript and square roots: a_{i+1}, x^{n1}, sqrt{1  x^2} and use "{}" as normal characters everywhere else. I prefer the last one. It allows us to use backtics for something else  like plain text: `number of bananas` N =  7 In my little toyparser I added support for fractions  so that we can do: .. math:: 1   beta x + 1 f(x) =  x^2  1 We should also have a way to do matrices. Perhaps something like: 1 :  : 0 a : 0 a : *M* = (.....)^1 = (.........) 0 : b : 1 0 : : b JJ 
From: Colin J. Williams <cjw@sy...>  20041213 13:59:46

Jens Jorgen Mortensen wrote: > David Goodger wrote: > >> I have interspersed specific comments below. My general comment is >> that I don"t think reST should invent a new math syntax. There are so >> many details and intricacies, it"s opening a huge can of worms. >> Instead, we should adopt one or more existing math syntaxes. I"d >> rather see an exploration and discussion of the available, existing >> syntaxes, available software, and output support options. >> >> There"s a practical reason for my position as well. Up to now, nobody >> has had the time or will to implement a proper "itex" or "tex" or >> "mathml" directive (except for Beni"s math hacks, but they"re hacks >> ;). I very much doubt that anyone will jump on this opportunity to >> do a lot more work for fewer rewards. >> >> You"re welcome to pursue this, of course. If you do, I hope you >> succeed. But I have neither the time nor the desire to do much for >> it. > > > The discussion sort of stopped here! Probably because nobody wants to > do the work. I am willing to do the work if we can agree on what to > do. I already implemented a parser for a syntax similar to that > proposed by Alan > (http://sourceforge.net/mailarchive/message.php?msg_id=10020457) some > time ago. > > The big question is: *Should* we have a reSTnative math syntax? If > yes, what should it look like? > > I like the proposal of Alan  I am a little in doubt concerning the use > of backtics for grouping. Grouping can be used for distinguishing: > > a_`i+1` and a_i + 1 > > Alternatives: > > * Use lack of whitespace as grouping: > > a_i+1 and a_i + 1 > > * Use "{}" for grouping (::math:`a_{i+1}`)  we loose "{" and "}" > as normal characters. > > * Use "{}" for grouping only for subscript, superscript and square > roots: > > a_{i+1}, x^{n1}, sqrt{1  x^2} > > and use "{}" as normal characters everywhere else. > > I prefer the last one. It allows us to use backtics for something > else  like plain text: > > `number of bananas` > N =  > 7 > > In my little toyparser I added support for fractions  so that we can > do: > > .. math:: > > 1 >   beta > x + 1 > f(x) =  > x^2  1 > > We should also have a way to do matrices. Perhaps something like: > > 1 : >  : 0 > a : 0 a : > *M* = (.....)^1 = (.........) > 0 : b : 1 > 0 : > : b > > JJ I believe that Alan Isaac made a later posting than the one quoted above. 18Nov04, he said: This is very close to what I had in mind! http://www1.chapman.edu/~jipsen/mathml/asciimath.html This provides an immediate solution for reST+math > XHTML and, if I understand correctly, a substantial part of the solution for reST+math > LaTeX and other formats. Cheers, Alan Isaac Someone else has designed the syntax, it looks neat. Colin W. 
From: Alan G Isaac <aisaac@am...>  20041213 19:03:35

> Jens Jorgen Mortensen wrote: >> The big question is: Should we have a reSTnative math syntax? If >> yes, what should it look like? ... >> In my little toyparser I added support for fractions On Mon, 13 Dec 2004, "Colin J. Williams" apparently wrote: > I believe that Alan Isaac made a later posting than the one quoted > above. 18Nov04, he said: >>> This is very close to what I had in mind! >>> http://www1.chapman.edu/~jipsen/mathml/asciimath.html As Colin says, someone has designed a workable syntax. In addition, the syntax is parsed by a pretty simple JavaScript, which has a php port: http://www.jcphysics.com/ASCIIMath/ I have argued that this workable syntax can therefore be ported with relative ease to Python and to reST. I have also volunteered to help test and improve code, but as a nonprogrammer I cannot envision the core architecture. This could be rapidly implemented and would give reST really nice math support with relative ease. Indeed, if reST would support this with nothing more than a 'math' role and directive that does nothing more than pass the ASCIIMathML raw plus delimiters at each end, we could support the LaTeX subsyntax of ASCIIMathML *immediately*! (In HTML documents a javascript link would have to be added; in the LaTeX documents *nothing* would have to be added.) There is a high reward  low cost opportunity here. I hope Jens will consider attacking this! My offer to help is still good. I note that Peter Jipsen (creator of ASCIIMathML) has also expressed modest interest. fwiw, Alan Isaac 
From: Jens Jorgen Mortensen <jensj@fy...>  20041215 09:38:49

Alan G Isaac wrote: > As Colin says, someone has designed a workable syntax. > In addition, the syntax is parsed by a pretty simple > JavaScript, which has a php port: > http://www.jcphysics.com/ASCIIMath/ > I have argued that this workable syntax can therefore > be ported with relative ease to Python and to reST. > I have also volunteered to help test and improve code, > but as a nonprogrammer I cannot envision the core > architecture. I did have a look at asciimath and it seems to be very nice work, but does it qualify to be the official reST math markup syntax? It is more readable than latex, but not much more. For now I would like to have a tex role and a tex directive. If I implement that for the latex and html writers (using MathML) would that have a chance of getting into the docutils code? JJ 
From: Alan G Isaac <aisaac@am...>  20041215 16:51:01

On Wed, 15 Dec 2004, Jens Jorgen Mortensen apparently wrote: > I did have a look at asciimath and it seems to be very > nice work, but does it qualify to be the official reST > math markup syntax? Unfortunately, the developers have not offered a clear vision of the desired scope of math for reST. I believe however that David Goodger suggested that there may be more than one implementation, ultimately. I love LaTeX, and I personally would be *fully* satisfied with the role and directive you describe, but I feel ASCIIMathML is a better choice for reST. Granted it is more limited than LaTeX, but it does a helluva lot. It is much more "natural" for the average person to read, and I think of readability as part of the reST goals. > It is more readable than latex, but not much more. I disagree. I think for a nonLaTeXuser, it is much more readable. And LaTeX users can use its LaTeX syntax, which covers quite a bit of ground *and* can easily be extended. > For now I would like to have a tex role and a tex > directive. If I implement that for the latex and html > writers (using MathML) would that have a chance of getting > into the docutils code? My guess is that the developers would be very open to this. I recommend choosing itex instead, so that you can get a running start with your project with itex2mml. The main difference is in array handling, where itex is structured to facilitate the conversion to MathML. hth, Alan Isaac PS I would happily participate in testing and (if appropriate, as I program only at the hobbyist level) in bug fixing. 
From: Colin J. Williams <cjw@sy...>  20041215 18:18:43

Alan G Isaac wrote: >On Wed, 15 Dec 2004, Jens Jorgen Mortensen apparently wrote: > > >>I did have a look at asciimath and it seems to be very >>nice work, but does it qualify to be the official reST >>math markup syntax? >> >> > >Unfortunately, the developers have not offered a clear >vision of the desired scope of math for reST. I believe >however that David Goodger suggested that there may be >more than one implementation, ultimately. > >I love LaTeX, and I personally would be *fully* satisfied with >the role and directive you describe, but I feel ASCIIMathML >is a better choice for reST. Granted it is more limited >than LaTeX, but it does a helluva lot. It is much more "natural" >for the average person to read, and I think of readability >as part of the reST goals. > > > >>It is more readable than latex, but not much more. >> >> > >I disagree. >I think for a nonLaTeXuser, it is much more readable. >And LaTeX users can use its LaTeX syntax, which covers >quite a bit of ground *and* can easily be extended. > > > As a nonLatex person, I'm inclined to agree that ASCIIMathML is much more comprehensible. It seems to meet my limited needs. To Jens I would ask what functionality is available with Latex math and is not provided by ASCIIMathML. Also, how significant are these features for the average user. Colin W. >>For now I would like to have a tex role and a tex >>directive. If I implement that for the latex and html >>writers (using MathML) would that have a chance of getting >>into the docutils code? >> >> > >My guess is that the developers would be very open to this. >I recommend choosing itex instead, so that you can >get a running start with your project with itex2mml. >The main difference is in array handling, where itex >is structured to facilitate the conversion to MathML. > >hth, >Alan Isaac > >PS I would happily participate in testing and (if appropriate, >as I program only at the hobbyist level) in bug fixing. > > > 
From: Colin J. Williams <cjw@sy...>  20041215 23:25:33

Alan G Isaac wrote: >On Wed, 15 Dec 2004, Jens Jorgen Mortensen apparently wrote: > > >>I did have a look at asciimath and it seems to be very >>nice work, but does it qualify to be the official reST >>math markup syntax? >> >> > >Unfortunately, the developers have not offered a clear >vision of the desired scope of math for reST. I believe >however that David Goodger suggested that there may be >more than one implementation, ultimately. > >I love LaTeX, and I personally would be *fully* satisfied with >the role and directive you describe, but I feel ASCIIMathML >is a better choice for reST. Granted it is more limited >than LaTeX, but it does a helluva lot. It is much more "natural" >for the average person to read, and I think of readability >as part of the reST goals. > > > >>It is more readable than latex, but not much more. >> >> > >I disagree. >I think for a nonLaTeXuser, it is much more readable. >And LaTeX users can use its LaTeX syntax, which covers >quite a bit of ground *and* can easily be extended. > > > As a nonLatex person, I'm inclined to agree that ASCIIMathML is much more comprehensible. It seems to meet my limited needs. To Jens I would ask what functionality is available with Latex math which is not provided by ASCIIMathML. Also, how significant are these features for the average user. Colin W. >>For now I would like to have a tex role and a tex >>directive. If I implement that for the latex and html >>writers (using MathML) would that have a chance of getting >>into the docutils code? >> >> > >My guess is that the developers would be very open to this. >I recommend choosing itex instead, so that you can >get a running start with your project with itex2mml. >The main difference is in array handling, where itex >is structured to facilitate the conversion to MathML. > >hth, >Alan Isaac > >PS I would happily participate in testing and (if appropriate, >as I program only at the hobbyist level) in bug fixing. > > > 
From: Jens Jorgen Mortensen <jensj@fy...>  20041216 08:19:27

Colin J. Williams wrote: > Alan G Isaac wrote: >=20 >> On Wed, 15 Dec 2004, Jens Jorgen Mortensen apparently wrote: >> =20 >> >>> I did have a look at asciimath and it seems to be very >>> nice work, but does it qualify to be the official reST >>> math markup syntax? >>> =20 >> >> >> Unfortunately, the developers have not offered a clear >> vision of the desired scope of math for reST. I believe >> however that David Goodger suggested that there may be >> more than one implementation, ultimately. So LaTeXmath could be one implementation. >> I love LaTeX, and I personally would be *fully* satisfied with >> the role and directive you describe, but I feel ASCIIMathML >> is a better choice for reST. Granted it is more limited >> than LaTeX, but it does a helluva lot. It is much more "natural" >> for the average person to read, and I think of readability as part of=20 >> the reST goals. I agree with you  we should have something more readable than LaTeX in=20 the long run, and ASCIIMathML is probably a good choice. But I would=20 like to start working on the LaTeX role and directive and gain some=20 experience with that first. > As a nonLatex person, I'm inclined to agree that ASCIIMathML is much > more comprehensible. >=20 > It seems to meet my limited needs. To Jens I would ask what > functionality is available with Latex math which is not provided by=20 > ASCIIMathML. Also, how significant are these features for the > average user. I don't see anything significant missing. > Colin W. >=20 >>> For now I would like to have a tex role and a tex >>> directive. If I implement that for the latex and html >>> writers (using MathML) would that have a chance of getting >>> into the docutils code? >>> =20 >> >> >> My guess is that the developers would be very open to this. >> I recommend choosing itex instead, so that you can >> get a running start with your project with itex2mml. >> The main difference is in array handling, where itex >> is structured to facilitate the conversion to MathML. Could you give an example of the difference between itex and LaTex=20 arrays? If we choose itex syntax then we will have to convert that to=20 LaTeX style in the LaTeX writer. >> >> hth, >> Alan Isaac >> >> PS I would happily participate in testing and (if appropriate, >> as I program only at the hobbyist level) in bug fixing. That would be great! Jens J=F8rgen 
From: Alan G Isaac <aisaac@am...>  20041217 03:43:50

>> Alan G Isaac wrote: >>> I believe >>> however that David Goodger suggested that there may be >>> more than one implementation, ultimately. On Thu, 16 Dec 2004, Jens Jorgen Mortensen apparently wrote: > So LaTeXmath could be one implementation. That is my understanding. > Could you give an example of the difference between itex and LaTex > arrays? If we choose itex syntax then we will have to convert that to > LaTeX style in the LaTeX writer. Start at http://pear.math.pitt.edu/mathzilla/itex2mmlItex.html Unfortunately some key links there are bad, but you will find that itex essentially extends WebTeX (in ways we can discuss as this moves forward). For WebTeX documentation, look here instead http://stuff.mit.edu/afs/athena/software/webeq/currenthome/docs/webtex/toc.html With arrays, itex follows the WebTeX conventions. Examples can be seen at http://stuff.mit.edu/afs/athena/software/webeq/currenthome/docs/webtex/examples/ Particularly relevant is option specification: http://stuff.mit.edu/afs/athena/software/webeq/currenthome/docs/webtex/wtxsec7.html#ARROPT In my opinion, translation to LaTeX is not challenging, but translation to MathML is. However the itex/WebTeX structure was chosen to facilitate translation to MathML, and itex2mml is open source and does the translation already (in part by using flex on a translation table from itex to MathML)! I believe that taking this route could be jointly profitable to improving the existing itex2mml and to extending it to translate an itex role and directive for reST purposes. So it is my belief that a lot of the MathML work has been done, if you speak C, and the LaTeX work is I believe relatively trivial. (Of course, the proof is in the pudding.) Translating WebTeX to LaTeX has already been given some thought as well, e.g., if you speak perl, http://www.lib.rpi.edu/dept/acs/rpinfo/filters/GChtml2latex/webtex2latex.tag makes a substantial start on this! hth, Alan Isaac 
From: Jens Jorgen Mortensen <jensj@fy...>  20041217 07:31:02

Alan G Isaac wrote: > So it is my belief that a lot of the MathML work has been > done, if you speak C, and the LaTeX work is I believe > relatively trivial. (Of course, the proof is in the > pudding.) Translating WebTeX to LaTeX has already been given > some thought as well, e.g., if you speak perl, > http://www.lib.rpi.edu/dept/acs/rpinfo/filters/GChtml2latex/webtex2late= x.tag=20 > makes a substantial start on this! I have written a parser that transforms LaTeX to MathML in Python  so=20 that is not a problem. It doesn't do all of LaTeX, but it's a start. I am going to ask some questions on the docutilsdevelop mailing list=20 (Alan: you are not on that list!). Jens J=F8rgen 
From: Felix Wiemann <Felix.W<iemann@gm...>  20041220 18:38:37

Jens Jorgen Mortensen wrote: > I did have a look at asciimath and it seems to be very nice work, but > does it qualify to be the official reST math markup syntax? Frankly, I do not think anything should become the official math syntax of reST, because then people would start endless discussions about how to improve the math support and why not choose $FANCY_MATH_SYNTAX for Docutils. > For now I would like to have a tex role and a tex directive. If I > implement that for the latex and html writers (using MathML) would > that have a chance of getting into the docutils code? Probably not, because that would mean favoring one math syntax over another. And math support probably isn't a particularly simple thing, so it might very well be too big and complex for a system like Docutils. I suggest you first publish it as a patch (or branch, as you like) and when there is some plugin support in Docutils, it can be turned into a plugin. (I'd be willing to help with that.) Please understand that my objections are against including math support in the Docutils core, not against math support in general. The math plugin(s) could get some wellvisible announcement on the Docutils homepage.  When replying to my email address, please ensure that the mail header contains 'Felix Wiemann'. http://www.ososo.de/ 
From: Marcelo Huerta <mghsm@sp...>  20041221 02:37:35

El 20/12/2004 a las 15:25, Felix Wiemann <Felix.Wiemann@...> dijo= , en su mensaje "[Docutilsusers] Re: math in reST": > Frankly, I do not think anything should become the official math sy= ntax > of reST, because then people would start endless discussions about = how > to improve the math support and why not choose $FANCY_MATH_SYNTAX f= or > Docutils. <fx: claps hands enthusiastically> +1 =20 o=3D< Marcelo >=3Do cancha. Chancha donde se practica un deporte. Del "Bichonario" (Gim=E9nez/Wright) 