First login?
Lost password?
 
 
 
mailing-list for TeXmacs Users

Text archives Help


Re: Some comments on new features and some ideas


Chronological Thread 
  • From: Joris van der Hoeven <address@hidden>
  • To: address@hidden
  • Cc: address@hidden
  • Subject: Re: Some comments on new features and some ideas
  • Date: Wed, 11 Sep 2002 15:58:38 +0200 (MET DST)

Thanks for raising these issues Alvaro.

> Cygwin TeXmacs
> --------------
> What did it happen with the project to include it in the main Cygwin
> distribution? That would do much good to TeXmacs. On my side, I have had
> problems when compiling with the latest cygwin. It complains when
> arriving at evaluate.cc from an struct.h found in /usr/include/libguile
> (sorry, no precise error description at hand, it's on a win XP machine
> now unavailable to me).

Unfortunately, Marciano Siniscalchi (in the CC),
who did the Cygwin port, lost his PC with Cygwin when
changing his job recently. Is there someone else who can
take a careful look at this issue and do the maintaining
until Marciano will be able to take this up again?

I also agree that it would be cool to include TeXmacs
in the main Cygwin distribution (if that is possible).
It would also be cool if someone is able to write
an automatic installer for Cygwin. In that case
we might sell CD's by Internet as a service.

> Presentation mode
> -----------------
> It works for me under KDE but crashes the application under fluxbox.
> Anyone could report about his/her favourite windowmanager?. It is a
> very nice feature, but it would be nicer if the keyboard worked for
> navigating within.

Stéphane promised us to correct this bug by finding the appropriate
WM extension proprerties (the problem is that X is a mess) for
setting the windows geometry and border. However, I suspect that
he has been very busy with Qt and Perl. Nevertheless, it would be
nice if Stéphane could take a couple of days to fix these issues
and/or explain to us how these WM extension properties work.

> I think also that the "reduction factor" way of
> calling "zoom" would welcome some thoughts.

Yes.

> Also an icon for RF would make switching faster.

Yes, but we have to be careful with room.

> Folding structures
> ------------------
> Folded is great but, why the empty bullet marker? There must be some
> reason why you have implemented it that way, Joris, what is it?

Folding is not yet well done. I have just started experimenting with it
(for some personal presentations), but this issue has to be worked through
more carefully after a series of other changes.

> When I
> thought about "folded" I was thinking about being able to fold/unfold by
> double-clicking a section title, but after an explanation from David on
> the demerits of recursive sections with regard to tree operations, I
> understand it must be done otherwise.

I would be interested by these explanations, because I also see some
merits of recursive sections...

> This folding thingy would be great to design a "note to other authors"
> environment.
> The folding environment raises again the debate of whether a
> supplementary area giving visual clues about the underlying structure of
> the document would be helpful or mostly annoying / superfluous. My
> example for folding is the Scite source code editor, which uses the
> margin to mark folded regions.

I agree; but this kind of markup is not our priority for
the next 6 months :^(

> Switch & Superpose
> ------------------
> I like to play with switch, although I have problems finding an
> use-case. I would welcome ideas, because it's so cool!. I give my own:
> the unoptimized version of an algorithm might be stored in one switch,
> the optimized in the next, the actual code in the next, and so on. This
> poses problems when producing an output document, because one might like
> to see all switches one after another. This option AFAIK doesn't exist
> now. It's very dynamic, but I see most users just putting the
> algorighms sequentially if no transition function is provided.

You might want to design a style file where the switch macro just does
what you want...

> I also see switch useful in demonstrating the use of styles in TeXmacs.
>
> What's the aim of superpose?. I mean, what does that resolve? I see it's
> a more general way to do strike-through, and it is funny indeed to
> superpose \sum with \cup, but are there other intended uses?

Graphics, but I do not have time to work on this now unfortunately.
It may also be useful for presentations in combination with switch:
make more and more things appear in a progressive way without
changing the position of what is already visible.

> For the documentation task I would like to be as informed as possible on
> these features, that's why I make so many questions...

Would you like to write some documentation on these issues?

> Variants
> --------
> They are great!!. Congrats, Joris, because they work amazingly well. It
> remains to be seen what is the performance in long documents, a
> preliminary test I made with an imported LaTeX document wasn't very
> encouraging on that side.

They are mainly used for smaller structures, like itemize<->enumerate,
equation<->equation*, theorem<->proposition, tabular<->block, etc.
For section titles it may be a bit slow.

I also foresee some enhancements in the future for
"upgrading and downgrading", but I don't promise a date.

> BTW, thanks for the improvements in the LaTeX
> input filter, Gwenael!.
>
> There's a slight bug I'd like to report in the variants.scm: "exercice"
> and "problem" are variants. The typo in "exercice" prevents the variant
> from using the exercise style.

Oh, I will have to check this.

> I'm not sure whether variants belong with TeXmacs or with the user's
> behavioursheet (document-dependent perhaps?) or if variants could be
> reported somehow on the stylesheet. I think this has to be thought of
> carefully.

It will be part of what I call a XEL (extensable editor language) or
DRD (data relations definition). These are extremely powerful concepts
which I will finally be able to implement now that we have version 1.0.
More on that later.

> Also I would very much welcome using variants for the paired bracketing
> scheme I suggested some time ago (a shortcut to enter both delimiters at
> the same time and have the cursor placed in the middle of them). It
> would be nice to have variants change the delimiter style then.

Yes, brackets might be part of the structure.
But where do we stop? Should + be a binary structure too?
There are basically two approaches for letting (), +, etc.
behave in a structured way:

1) No structure in the document, but the incorporation of
powerful parsing algorithms in order to *compute*
the structure as a function of a grammar.

2) What would be the resulting parse tree of 1 is used
as the structure of the document itself
(so (x,...,z) is a (2n+1)-ary macro, + a binary macro, etc.).

What do you think about this Stéphane? This is close to litterate
programming. The first approach makes editing simpler,
but manipulating the structure harder, while the second approach
does precisely the inverse.

This is also interesting w.r.t. a recent discussion with David of
the potentitial use of cursor positions in intermediate steps
of the rewriting procedure. Indeed, we might do the parsing in 1
by means of a rewriter and try to use the structure of the parse tree
for manipulating the "structure" of the original source.
What do you think about this David?

> Ultra-newbies
> -------------
> This is an experiment I am conducting with my mother (she's writing a
> book on spanish geography), a person new to computers. I want TeXmacs to
> pass the newbieuser test, because it performs so well already in many
> respects that it would be a pity not to care about these users.
>
> Specificly, my mother has problems with 1) detached selection

That is?

> 2) the -> arrow not wrapping lines,

I agree. Time, time, ...

> and 3) the line-movement commands.

What do you mean?

> She also
> crawls with the open file dialog because there's no distinct indication
> of which is a directory listing and which area is a file listing.

Isn't that pritty clear from the fact that the directory listing
always contains ..?

> On the bright side, once explained the nesting of environments in
> TeXmacs it has become easy for her to do her complex lists of lists of
> lists. She also has benefitted from the C-tab variant scheme, and it's
> been very encouraging the use of A-backspace to remove enclosing
> structure. So yes, I definitely think the new editing paradigm can be
> taught (although more problems are to be expected, she's only
> beginning).

Yes, please keep us informed about this learning process,
which is indeed very interesting.

> Mathematical operators
> ----------------------
>
> I have not yet settled my opinion about implicit operator syntax (type
> more than one letter within a math environment and you get an operator).
> There's people around arguing against on the case of "explicit is good"
> (the python credo, somewhere in the internet). I also don't like having
> to provide explicit spacing myself (although I ignore some feature for
> this perhaps). In any case, I think any operator should inmediatly
> create a child for its argument. Then, when you are inside log's
> argument (marked by the customary blue box) you could C-tab to obtain
> delimiter variants, including whitespace.
>
> The same principle could be applied for integral signs, sums, etc.

I agree that there is room for divergent opinions here and
I certainly intend to do some things about this in the direction
of what I said above about (), +, etc.

> Structured boxes and their colouring
> ------------------------------------
>
> I support the idea of colouring differently (but not too differently)
> the inner structure, because that signals clearly what will operations
> such as A-backspace apply to.

I agree with this. In fact I have been thinking about something even more
subtle in relation with structured cursor movements and editing:
Being able to let this "innermost structure colored box"
not be the real innermost structure, and also different from
the current selection. Structured editing commands
(like variants or structured movement) would than
be relative to this "intermediate structure colored box".

> I would also like "word" to be the lowest
> level unit for "strong" "emphasis" and so on in absence of an explicit
> selection. It seems people emphasizes 99% of the time whole words or
> groups of them, not isolated characters.

I'm not sure; if I type "strong", I do not necessarily want
the current word to appear strongly (in more than 1% of the cases).
You can always select the outmost word by "C-space C-space" and
apply strong to it...

> PS. regarding translations, I'm willing to coordinate the spanish
> subproject, with Offray and Pablo. Everybody else is invited.

Great. For the version 1.0.1 I want to switch to the online documentation
mechanism on Savannah (by providing tools to assemble complete
documentation from the fragments). Did you manage to use the CVS?
I hope to setup something better next year, but for the moment
it seems to be an acceptable way for making modifications.
You might also send me tarballs of new versions if you prefer.

> We'll have
> though to structure and complete a plan to adress in order the most
> important issues. I'll write something up in the wiki at the end of the
> month.

OK. Don't hesitate to give us hints about how we should improve
the tmdoc DTD (and style) either.

> I will also update the menu translations now their names seem stabilized
> when 1.0.0.16 comes out.

I think that there might be a few changes left, but not too many.

Best wishes, Joris


Archive powered by MHonArc 2.6.19.

Top of page