[Wikipedia-l] Parsing TeX

[Toby Bartels]

Jan Hidders wrote:

>Toby Bartels wrote:

>>But now should I understand that you won't normally use it in an
>>expression like [$x^2y = z_1$]; you'd write [$x^2$][$y$] = [$z_1$], and
>>save the full power of LaTeX for the fancy stuff? That would keep people
>>from trying [$x% = x/100$], mostly.

>Ah, isn't communication wonderful if it works. :-) Actually I would even
>expect people to write  [$x$]<sup>2</sup>[$y$]. However, if I'm honest I
>have to admit that my fingers would be itching to change that to [$x^2y =
>z_1$].

If people write [$x$]<sup>2</sup>[$y$] = [$z$]<sub>1</sub>,
then we really gain nothing from using the LaTeX syntax.
OTOH, if you change it to [$x^2y = z_1$], then people that copy you
will be likely to write [$x% = x/100$], even without reading
[[Wikipedia:How does one edit a math page]] or whatever.
So we have a difference between simple math that doesn't need LaTeX
and complicated math that people need to know LaTeX to use.
And these will be written in two different wiki styles as well,
just using the same symbol [$...$] (or whatever).

>>If [$...$] calls LaTeX, then I hope that you mention it ''somewhere''.
>>I understand that you don't have to mention it at the ''beginning''.

>Well, having thought about it a bit more, maybe I *would* mention it at the
>beginning. If [$ .. $] means that its contents is going to be interpreted as
>LaTeX markup then that is what the first sentence of its description should
>say. But I would follow that with saying that for simple mathematics you
>only have to know [$x_1$] and [$y^k$] means.

Perhaps it would be more profitable just to have a system
that handles scripts and not the rest of LaTeX, for ordinary use
(that is, use without the \sum s and \mapsto s of this world).
This thought is expanded upon below.

>>I also am speaking about ''if'' we introduce a markup for LaTeX;
>>but my argument is that it shouldn't make simple things any harder,
>>or introduce surprises into simple things (like weird behavior for "%").

>I agree, and that is one of the reasons that I now think that we should
>mention LaTeX right from the beginning. People then would know that "weird"
>stuff might happen if they are going to put more than just variable
>names between [$ and $].

Still, as you taught me, many people will just copy this notation,
and their usage won't satisfy the [[Principle of least astonishment]] ^_^.

>>I remember these, and I agree that there's no point in rehashing them.
>>It's the comment that <var> was never intended for mathematical variables
>>that caught my eye.  I'd like to hear more about that, if you want, either
>>here or elsewhere.

>Well, I didn't consider that an important argument because it is what
>browser are actually doing with it at the moment that matters. Anyway.

[snipped for brevity; see Jan's last post]

All this makes me wonder if perhaps it should be in <i> anyway.
You'd probably agree with this too; and most browsers don't distinguish
<i> from <var> (although lynx does, or can, as I've mentioned before).

----

Maybe we should do this instead:

Develop a wiki syntax for subscripts and superscripts
that can be used for math ''or'' other things, such as chemistry.
(My best suggestion now is "_{...}" and "^{...}".
For simplicity in the algorithm, braces would ''not'' be optional,
which would also spare random underscores in other contexts.
And perhaps it would be best to use something other than braces too.
But these details aren't central to what I'm talking about now.)

I should give up on this <var> crap and use <i> instead.
(You can continue to try to convince me to use '' and '''
in every situation instead of <i> and <b>, while this is going on.)

The above can be implemented now (or at least soon);
meanwhile, we all work on a LaTeX system intended only
for complicated (displayed?) mathematical expressions,
and which, like tables, will be an advanced feature
(although as accessible as we can manage to make it)
that people won't copy without knowing that it's special.
This LaTeX (or similar) won't be used for most math expressions,
but we won't have two systems for math expressions either,
because there ''is'' no special syntax for most math!
(The <sup> and <i> markup would work for anything, not just math.)


-- Toby Bartels
   <toby+wikipedia-l at math.ucr.edu>