Re: [m17n-list] math unicode

[Urs Holzer]

विश्वासो वासुकेयः (Vishvas Vasuki) wrote:
> [...] The lack of an easy way (not requiring a
> markup language) to specify superscripts and subscripts is sad - If
> there were unicode-characters like: START_SUPERSCRIPT,
> END_SUPERSCRIPT and so on, one wouldnt have to rely on html-like
> syntax. What do you think? Let's take this offline if you might be
> interested in pushing for its inclusion in the unicode standard..

(Posting to mailing list, since this could interest everyone. And see 
last paragraph.)

I think this is not a good idea for two reasons:

1. I don't know much about Unicode, but in my opinion, reading and 
writing Unicode should be kept as simple as possible. After all, basic 
text editors should be able to handle Unicode completely and entirely. 
With START/END characters, you are able to encode a whole tree-
structure. This would open the possibility that START/END characters do 
not match. It would mean that text editors would have keep track of 
them. They would have to look back in a file and count all the START/END 
characters up to the point the user wants to edit. I firmly believe that 
text is not a tree. I consider super- and subscripts not to be text but 
instead rules how to lay out text on a surface. This is what markup is 
for.

2. Superscripts and subscripts are far more complicated in mathematics 
than you might think. First they are attached to whole expressions, not 
only to a single character (e.g. (a+b)^2 is attached to the expression 
(a+b) as a whole). There are upper and lower scripts before as well as 
after an expression. Multiple superscripts are not unusual. I even bet 
that there are many stylistic differences employed for superscripts in 
various books. It should be possible to use styling rules for every 
script independently. This whole complexity means that we are basically 
stuck with markup. 

I admit that MathML is very verbose. If you want to keep your sanity, 
you are adviced to use a MathML editor. Another possibility is two write 
LaTeX or some other less verbose markup and then run it through a 
converter. MathML's advantage is that it explicitely encodes the 
structure of a formula. This information is essential for displaying, 
converting to speech, copy-pasting into other applications, and also for 
styling.

A completely different idea (but rather a hack) is to let m17n expand 
some special key combinations to MathML tags (#s for "<msub>", !s for 
"</msub>", #S for "<msup>", !S for "</msup>" or whatever).

Sorry for the rather long answer.
Greetings
Urs