Re: [Texmacs-dev] Re: Margins and export filters

[Nix N. Nix]

On Wed, 2003-03-12 at 14:45, Joris van der Hoeven wrote:
> > > Don't default style parameter values /become/ part of the structure
> > > as soon as the buffer comes into existence ? After all, if I open up
> > > TeXmacs and just start typing my, say, letter, then print it, won't
> > > I get default margins and a default document structure ?
> > 
> > I agree. And I think the best way to get out of the dilemma is to pass
> > the converter a typesetter loaded with the document to convert. Then
> > you can get whatever metadata you want.
> 
> No: if the margins are not explicitly set by the user, then they are not
> part of the structure, so LaTeX may chose other defaults than TeXmacs.
> 
> I also recall that TeXmacs does not aim to reproduce or approximate
> the LaTeX layout, even though we might add some facilities to this aim.
> For instance, we might add a configuration option for including
> default style parameters during the export. However, I do not think
> that this would be a good default option.

Keep the end user in mind during all this.  As an end user, I tend to
think of a converter as something that will /exactly/ reproduce what I
see to a different format.  I will, however, understand minor
differences, due to non-overlapping feature sets of the various formats
(e.g. PNG -> JPEG, i.e., lossy conversion).  However, I /do/ expect the
two documents to be very, very close.

Now, with TeXmacs and LaTeX, there is an added complication:  Not only
do we have to make the document "look" very, very close, to keep the
user happy, but we must also preserve it structurally as much as
possible, to keep the user happy.  As far as I know LaTeX, it is
flexible enough for one to specify very precise placement of text - now,
I'm not suggesting we treat LaTeX like a printer, but rather, we can
reproduce the various environments TeXmacs offers /very accurately/ in
LaTeX.  We could therefore (and we do, to a large extent, AFAIK),
declare environments in LaTeX that exactly (or, as exactly as partially
overlapping feature sets allow) reproduce the ones in TeXmacs. This
would keep the user happy - on both counts.  I know there are some
things LaTeX just outright doesn't support (like most wide-unders) so,
in those cases, we must write it off as a non-overlapping-features
issue.

As far as margins are concerned, I'm afraid I must agree with David.  A
user expects a document to have margins, without explicitly having to
specify them.  The document was created in TeXmacs, therefore TeXmacs
defaults will be reflected in it, whatever form it may take (LaTeX,
HTML, Scheme, PS, PDF, whatever).  If you take that together with the
user attitude toward converters that I have described (perhaps
improperly), you will find it inevitable to specify the margins
immediately upon document creation.

Also, you have already made this compromise in the case of matching
brackets:  You have added information the user hasn't explicitly
specified in order to make the LaTeX document a more accurate replica of
the TeXmacs document - well, in this case it was necessary for the
correctness of the document, and not just a matter of philosophy, so
it's a rather extreme example.

I think we should assume that the end user knows nothing about LaTeX,
except for

% latex filename.tex
% xdvi filename.dvi

At that point, given the current conversion, the user will likely say:
"WTF ?!  It looks totally different !" - than, say, the PS preview.

Another idea (inspired by GIMP):  When exporting to a certain format,
the GIMP always pops a dialog box with the various format features and,
basically, what aspects of the document to save (given that the format
supports them), and how to save it.  Now, I know your ideology is to
have as few popups as possible, but this may be a time to ask the user
whether she wants to export the margins and, I'm sure, other stuff that
we could cram into such a dialog box.

Background:  Some of these thoughts are not mine.  The people around me
use TeXmacs and they want to write their papers in it.  They ask me,
after having produced their LaTeX output, why it's not as seen on the
screen, and I have to tell them that the converter is not yet perfected.

Consequently, I think the LaTeX converter should be taken to the point
where LaTeX itself is the limiting factor in every situation.  This is
not yet the case in every situation.  Until recently, table borders were
handled really poorly by the converter.  Now, LaTeX is the limiting
factor, in that it does not allow borders to be specified for each cell
(AFAIK) - so, now "the ball is in LaTeX' court".  I think margins are a
similar issue.  Another issue is 2-column format, which is very
important for papers.  Formatting a chunk of text as 2 columns is not
reflected in LaTeX.  Margins, IMHO, are another issue, as is
"make-title", and the list goes on.  However, all these things can be
mapped into LaTeX both visually /and/ structurally and, if not, well,
"it's LaTeX' fault".