On Thu, 16 Mar 2000, Tuukka Toivonen wrote:
>
> On Wed, 15 Mar 2000, Uwe Amann wrote:
>
> > While I agree with you, that a LaTeX parser would be very hard to
> > implement, why not try a different solution. I will try to describe
>
> I still think that an heuristic algorithm would do moderately good
> work and it would not be that hard, actually, to implement. It couldn't
> guess always correctly the Maxima formula from tex/lyx representation, but
> well enough for most uses. Lyx might have some (optional) hinting
> mechanism, tough.
I think the problem with a mostly correct algorithm is that in
mathematics a mostly correct solution could well be totally wrong.
So user interaction would always be required. Then there has to
be a way that the user can check not only the result, but also
what Maxima formula LyX generated. But if the Maxima formula is
wrong what should the user do? Perhaps LyX could recognice some
special cases, for example with subscripts and present the
user with a list of possible Maxima-constructs, but if it is a
case not handled in the <special case list> the user will have
to edit it manually.
Stop. If you meant by hinting mechanism something like this above,
then your idea isn't too different from mine.
Let's say, we have a little converter from latex for Maxima. This will
take care of most things like conversion of special identifiers
(pi, e,i etc.), and some other things. The result is then displayed
for the user in order to correct it. Then it is sent to Maxima.
If the user doesn't trust the converter and knows he wants to evaluate
the equation, then he could enter it in Maxima s syntax by hand.
(that would be me ;-) )
What I think is important, is that we get a way, the advanced user
can be shure that his calculations give the same results,as if entered
into the computer algebra system directly. If we add a converter, that
does most conversions that are unambiguos we gain also the benefit, that
the beginner can also use the cas features for simple calculations.
I hope I understood you right. I will try to paraphrase this over
the weekend a little bit clearer and then perhaps we could talk with
someone of the devteam of LyX what they think about your idea.
>
> > that preserves it. Why not create an input mode for LyX, that allows
> > to enter commands for maple, which are stored as plain text.
> > This is then passed over to maple (with an added latex(...) )
> > to evaluate, and the result can easily be displayed with the
> > capabilities of LyX.
>
> I have to disagee about the idea as exactly presented.
>
> > This would have the advantage that you won�t have to leave LyX
> > to perform computations, and that the commands which led to
>
> True, but...
>
> > The only disadvantage I can see with using this method
> > would be that there are two different syntax standards for
> > mathematical input in LyX. While this could be confusing for
> > new users of LyX, it should present no problem because anyone
> > who will use this feature will probably know maple and latex
> > syntax.
>
> IMO, this is huge disadvantage, huge enough to defeat the whole
> purpose. The big (?) success of LyX is that it's possible to use LaTeX
> *without knowing anything about tex/latex*. Knowing tex *will* help -- but
> for basic usage, it is *not* necessary.
>
> The same principle should go for using Macsyma/Maxima. The user should be
> able to write the equations just normally and then optionally evaluate
> them transparently with a mathematics program. Even if the user knows
> Maple/Macsyma, he might not see beforehand, when starting to write the
> equations, then he would like to solve it automatically. The equation
> might look simple at first.
>
> > not too proprietary syntax.Using Mathematica would require
> > some additional work, because Mathematica�s latex output uses
> > special macros for many things. (At least the Version I am using)
>
> Well, if the user has money for Mathematica and willingness to use
> proprietary programs he could as well use the program mentioned --
> scientific workplace or what was its name -- for his purposes.
Here I have to strongly disagree with you ;-)
While I have access to both maple and Mathematica on different
machines ) getting a licence for scientific workplace is sadly out
of question (for now). I didn t use mathematica as example for
a program that should be used, just as some programm I am
familiar with. If support for such programs should be added
for LyX then it should of course use non-proprietary programs.
I think if support for one program is added there should be
added support for as many programs as possible (think of
parallel evaluation of expressions in several different systems and
the possibility to compare the results).
>
> > I would be glad if some LyX-Developer could tell me if this
> > additions to LyX are possible.
>
> Everything is possible with *really* free (in the GNU sense) programs like
> LyX -- the question is, which is the most sane for (1) developers and
> (2) users.
I have again to say, that my ideas represent my own likes and dislikes.
While I am willing to sacrifice the direct wysiwym-representation of
the mathematical equations I input, for the benefit of having the
possibility to evaluate them, I understand that many will disagree.
Greetings
Uwe
