Hacker News Reader

Comment by tffgg | Hacker News Reader

I know it's unrealistic, but I wish someone would reinvent Latex. I don't really want anyone to learn it. It's sad it seems kind of the best tools for advanced use out there...

View on HN

It's not difficult to learn at a base level. The real difficulty in learning LaTeX is becoming comfortable with handing the keys over when it comes to presentation — WYSIWYG grabs people easily, and, once taken in, it is tremendously difficult to break them of their urge to fiddle with things. In many ways it is good that it takes some doing to do more than the basics — nice output for basic documents is almost a guarantee.

I second that. I was able to re-write my CV which was in Word to LaTeX in about 3 hours with help of online resources (no prior LaTeX experience). The output is high-quality, consistent and looks professional.

I didn't get into a lot of advanced stuff - just defined a few "newcolumntype"/"newcommand" and used them everywhere but that freed me from formatting worries and let me focus on the content.

On the other hand less common features are often so difficult to learn.

For me the worst is tikz. I know it’s not exactly latex. But I wished it was easier to use. The output is so good, but it’s so damn hard to learn.

What – you did not enjoy reading the 1200+ page manual?

But yes, I know what you mean. I use the cargo cult method for writing TikZ code myself. Look for examples of people doing something similar, then poke at the code until it does what I want. I am slowly developing some glimmer of understanding, though. I figure in another 100 years' time, I'll be a TikZ wizard at this rate.

Tikz was so frustrating that I ended up drawing my stuff in an SVG editor and importing a high-resolution export from there. I know it's not as cool or scalable as a proper Tikz diagram, but if you have any value on your time, then I just do not think that Tikz gives you enough to be worth it, at least for relatively simple drawings.

Having something like ktikz (or qtikz) helps:

https://github.com/fhackenberger/ktikz

However, it is not quite as helpful as latexdraw was for pstricks.

That said, I am not sure how much of that difficulty is due to tools lacking and how much is due to "describe vector graphics as code" being a hard problem.

Take a look at TeXmacs (http://www.texmacs.org) (does not share code with TeX, only the name recalls it!). It is at the same time structured and WYSIWYG (having accessible text source), completely programmable (user-defined native macros and Scheme) and has a high quality typography (includes a global page-breaking algorithm). There is a helpful mailing list (info at http://texmacs.org/tmweb/home/ml.en.html) and a forum (http://forum.texmacs.cn/).

Problem with TeXmacs that they stuck with the old technology too - they use[1] ancient Guile and that prevents them to use JIT and in turn prevents[2] distributions to update Guile itself.

[1] https://github.com/texmacs/devel/issues/9

[2] https://bugzilla.redhat.com/show_bug.cgi?id=1828124

I agree that some communication from the TeXmacs developers to the Fedora ones would be helpful (I documented myself, Fedora is using Guile 2 and still supporting 1.8 with a compatibility package). (More about the Scheme in TeXmacs later). Said this, I think the angle from which you are looking at the software is too narrow. A better angle is to look at the realized potential of the TeXmacs document preparation system: it does everything one could wish for. With "everything" I am exaggerating a little :-)

More in detail (I refer to [1] for a complete discussion):

1) It makes structured editing possible (and that is its main mode of operation). So if you want to mark your text with "semantic" markup you can do that, and in fact that is what using the software leads you to do.

2) It is controllable via programs

3) It has advanced typography (including mathematical formulae at the necessary level for professional mathematicians, as far as I can see), and it is pleasant and easy to work with: you can concentrate on writing, letting the program take care of the visual appearance and you do not need to interpret the sources while you are editing, to figure out where is the text that you want to edit

4) It has user-definable keyboard shortcuts that further help entering structured text

... and I feel it is enough for a short list of the features.

Back to Scheme. The development team is debating on which Scheme to use for the future of TeXmacs, and the debate is taking time. Here are some experiments [2],[3]. The current software (with Guile 1.8) works smoothly, with full functionality. For Linux installation, I chose the static binary they provide ... because on Ubuntu I don't have Guile 1.8 ;-)

[1] http://cahiers.gutenberg.eu.org/cg-bin/article/CG_2001___39-...

[2] https://github.com/mgubi/texmacs/tree/guile3

[3] https://github.com/mgubi/texmacs/tree/s7

In addition to the ones I have already given, here is another source of information about TeXmacs: https://texmacs.github.io/notes/docs/main.html

After this post, the developers became aware that there is an open issue for that and responded at https://bugzilla.redhat.com/show_bug.cgi?id=1828124

I hope that will help to bring the development forward.

Thanks, that's precisely I hoped for.

Reinventing LaTeX can be done. The huge problem is to reinvent the ecosystem of packages. This would take several years, if not decades.

The reinvention is called ConTeXt.

LyX is an excellent WYSIWYM GUI for it.

This should probably be the first option for anyone who wants to produce a LaTeX document without learning the language, for some reason. It has support for bibtex, drawings, and equations.

there was a project but it seems abandoned

https://github.com/patoline/patoline

Reply on news.ycombinator.com