On Mon, May 03, 2021 at 10:32:31PM +0200, Ralf Hemmecke wrote:
> > I can see that the current proposal might be undesirable if the same
> > worksheet is to be presented in another format, e.g. a worksheet
> > prepared in jfricas but later you want to use it in TeXmacs.
>
> OK, I as a user prepare some cool stuff in jfricas and then I have some
> work to make it look nice in Texmacs. Why is that a problem that FriCAS
> should deal with?
>
> If you aim at using a worksheet in both jfricas and texmacs, then use
> unicode. If, however, you solely use the jfricas interface and want the
> computed result for easy inclusion in a slide or other LaTeX document,
> then use my proposal.
If nobody uses your code, then problem is greatly limited, but
not gone. But to use your code people have to embed LaTeX
construct into things that should be pure math. You create
completely unnecessary fragmentation: to get what that want
in your proposal users have to create LaTeX-specific code,
which will fail with other formatters, or ignore your
feature. I could understand it if it was impossible to do
better. But there are _clearly_ better alternatives.
First, formatter may replace Unicode characters by appropriate
control sequences. AFAIK Unicode characters works fine in MathML
and there is support in Texmacs (I am unable to test it, but
is right way to go). Currently Unicode characters works also
on most terminals, so in text modes. So natural thing is to
make sure that they work with LaTeX. You say that there
are troubles on LaTeX side. In such case it is reasonable
to do substitution in the formatter: for supported characters
replace Unicode by appropriate LaTeX control sequences.
You write about "cool stuff", however need to mess with
LaTeX control sequences is very un-cool and rather
make impression of immature system that can not do
things properly.
There is also question of documentation: without documentation
such feature is much less useful. If it is documented, then
there will be confusion if we remove it (obsolete info tends
to have long life on the web).
Another things is that code that we offer should be a good
example. Unfortunatly, beginers frequently fall into trap
of "getting something to work" and follow bad examples.
In your proposal users are encouraged to follow bad
coding pattern and this patten is visible (as opposed
to something encapsulated in it own ghetto).
More generaly, FriCAS strength is its principled design.
In short term, by breaking principles one can frequently
make gains, this is called technical debt. But like
real debt, we either have to pay it back (clean up
things) or face insolvence. Technical debt means that
there will be increased cost of further developement
and IMO in most cases is wrong choice.
--
Waldek Hebisch