mgubi

I write all my papers with it since few years, and my lecture notes, and my talks. The papers are converted to LaTeX when sending to the publisher. Several of my PhD students used it to write their thesis. Also some Master students.

Some pointers here: Papers: - https://arxiv.org/abs/2004.09637 - https://arxiv.org/abs/2112.05562 - https://arxiv.org/abs/2107.12115 Lecture notes: - https://www.iam.uni-bonn.de/abteilung-gubinelli/teaching/sto... Online lectures with TeXmacs: - https://www.youtube.com/watch?v=FL50D2PpswE&t=4990s

We seldom choose our names, ourselves. The name was chosen early on to pay due respect to the two main sources of inspiration: the TeX system and the Emacs editor. TeXmacs takes the good features of both, but just because I have the same name as your parents it does not mean I'm the same kind of person. A family is not a clone hierarchy. One can argue that Javascript also has the wrong name, there are reason for this, but usually this does not stop people to appreciate the good sides.

Graphics is different since usually the graphics primitives are the common substrate of all graphics languages, so the interpreter can just be adapted to a different backend. For more structured content this is not possible: the natural output of TeX is a sequence of graphics boxes for the glyphs, or something like a DVI format. It is too late to retain semantic informations, e.g. about mathematical formula. Ask anybody which writes LaTeX/HTML converters for example. TeXmacs does a good job to interpret "usual" LaTeX files but you cannot expect to be able to interpret arbitrary files and still retain semantic informations, e.g.: sectioning, references, emphasised texts, theorem statements, change of languages (e.g. from English to C++ to Math...) etc...

It is a bit discomforting seeing that discussion in hackernews just remain at the level of "bad choice of name, btw!" without no technical insights. I would have expected more sensible reactions. TeX and Emacs were initial inspirations to TeXmacs (which has been written in the '90). As such they indeed inform some of the design choices, that said TeXmacs do not conform to them. The Emacs legacy is however very clear: all the editor behaviour is re-programmable in Scheme, from keybindings to menus, to the graphics editor (see here a recent example http://forum.texmacs.cn/t/why-not-draw-circle-by-center-and-...). As for the TeX legacy, TeXmacs actually improves on it by defining a document format while retaining the power of creating user defined macros, it also allows scheme code to be called from macros, since its beginning, 20+ years before LuaTeX. An overview of TeXmacs design can be found here: https://texmacs.github.io/notes/docs/overview.html . I think people nowadays are more used to programs which are epsilon changes over existing systems, repackaged with brand new names. That is why TeXmacs could be puzzling at first. Does not fit usual categories. Maybe this says something of the current trends in (free)software design.

You do not know what are you talking about clearly. What is "this powerhosrse difference"? What in your opinion is possible with one tool but not with the other?

Well, I guess that if you use "Python" you expect to really have to deal with a snake then.

No, document formats are not turing complete. HTML for example can be parsed without any problem. However take this TeX snippet "\foo [bar]" and tell me if "bar" is an argument of "foo" or not, without running the macro (which maybe you do not have. This is the basic problem any TeX conversion program encounters. TeX is not a document format but a programming language. The only thing you can do with a TeX program is to run it. On the other hand an HTML document can be parsed without knowing anything about its content.

Thanks for the detailed description. I understand better the problem. I'm not very accustomed to other scripts so I would need to look into this with more details. But at least now I understand what HarfBuzz is about :)

I've looked a bit around and in particular in the poppler library code to see how it works, they use QRawFont in the Qt backend to exploit subpixel antialiasing and native font technologies for each platform. We could do the same, essentially replace our glyph rendering code with something similar to their code. In this way we will make TeXmacs rendering similar to that of Okular, for example.

I'm looking right now into better OpenType support. Would you care to specify more in detail what is your use case and the problems you encounter?

Do you have some hints on where to look to see a good implementation of a text rendering engine implementing those features? (Browsers? which?)

The converter (e.g. to HTML or LaTeX) do not require strictly all the program but nobody took the pain to extract it and make it independent of the GUI layer. This could be a nice side project. We are also thinking to implement a facility to use TeXmacs as markdown editor. My personal experience is however that you want to create your content in TeXmacs and then eventually export to other formats, e.g. markdown or HTML to embed in blogs, etc..

I would be interested in improving (or at least experimenting) in this direction. Do you have any hint on what to do, or what was the problem when you tried? As with PDF we are not bound with a specific rendering pathway in the code, there could be the possibility to improve the Qt rendered for, at least, those fonts which are supported via Freetype.

In my experience pandoc is not better than TeXmacs in converting LaTeX. The problem, as vdhoeven stressed, is that LaTeX is not a document format which can be converted reliably in something else. Usually I just convert to LaTeX the last version of the document, e.g. before submitting to journal. In this way I always dispose of some alternative storage formats: TeXmacs, PDF, LaTeX and sometimes HTML if needed. The LaTeX produced by TeXmacs uses few packages, which also means that it can be easily maintained.

Well, is more an opportunity to move forward. You just have to consider LaTeX as another output format, not really for human consumption, similar to PDF and HTML. Something you do not want to edit by hand. Many people are anyway moving in this direction using Markdown+pandoc and similar solutions. TeXmacs is another possibility in this direction which allows you to concentrate on the contents, including structure, and not on the programming of the content.

That is always tricky. HTML is a presentation language in many ways since the use of tricks or inline CSS is not forbidden. So the problem of interaction with it is similar with the problem of interaction with LaTeX (which however adds on top the macro problem). We are doing research on conservative conversion from LaTeX and probably similar techniques could be used for HTML. However I think would be more meaningful to aim to Markdown or some restricted versions of HTML as common format, since this will is easier to map reasonably well to the TeXmacs document format. I wonder if tools like PanDoc have these bidirectional conversion capabilities themselves. We are planning to have a conservative conversion for Markdown documents (meaning that local edits in TeXmacs only change locally the original Markdown document)

Yes it does. I didn't got the phrasing right. Sorry! Anyway my comment is valid, I was not really sure if I should keep the recursive picture or not. But I didn't resisted...

Julia is a lot of fun.

http://forum.texmacs.cn/t/julia-programming-language/115/27

I do not think so. I understood you didn't liked the Figure comment (and maybe the figure itself)?

I found both links you suggest very interesting and impressive. But again part of my objection is that I do not want my screen divided in two, there is no need in 2020 to stick to this old tradition. I like more to be able to do this:

https://www.youtube.com/watch?v=ILiL4m61BLw&t=4020s

I've used Emacs+AucTeX for long time and I agree with you that it is an excellent toolset. It is simply that with certain tools you will never ever even think to do something because it is just these 5 sec (too much) away. Like using colors in a table, or making algebraic manipulations directly on the documents on complex formulas.

The tools we use shape what we do, subtly. If it takes you more than 5 min to make a drawing, maybe you take the decision not to do one. If takes more thant 30 sec to make a table, then same: maybe you get along without. In any case the burden of the technology will impose on you to get out of your writing focus to care about technical matters (open another applications, look into stackexchange...). I wrote that blog article to make a point for all those users who care about the quality of the interaction with the writing tool. I understand very well other tools allow more flexibility in some respects or wider compatibility, but in my personal experience none of the above mentioned tools can replace TeXmacs for creative technical writing or even for fast note taking.

If you compare TeXmacs and LyX then you miss the point of my blog article. I've tried to carefully explain the particular nature of TeXmacs including some careful UI design choices which allows the UI to be out of your way and out of your screen. Ideally, there should be no button on the screen (look at Emacs, no toolbar), the free space on the screen has to be allocated for the document, with the right information popping up when needed. Moreover I do not want to see an approximation of the final result, that is not good enough (and we are in 2020) It is not a problem with LyX itself: if you stick to 100% compatibility with LaTeX then there is no much more you can do.

TeXmacs is as extensible as Emacs, being based on the same principles: having a full-fledged scheme interpreter to modify the behaviour of the UI and to programmatically edit the documents. You can even have Emacs-like keybindings :). What Emacs lacks and TeXmacs has is a sophisticate typesetting engine (i.e. something like TeX) to render complex documents. What Emacs has and TeXmacs lacks is a large user base and dedicated people which develop extensions, but it has a small number of core developers which maintain and use it everyday. It has ~20 years of development history.

I'm sorry you do not like it. Actually is borderline for me too. But I didn't resisted...

> I bet this article itself was written in texmacs and exported to html.

you won. Actually as I wrote in another comment all the TeXmacs website, the blog in which the article appears, the web page of the main developer and some other websites are completely generated within TeXmacs with no assistance whatsoever from external static site generators. The content can be edited directly in TeXmacs and the hyperlinks works connecting the TeXmacs documents, so you can navigate and edit the website from within the program. This workflow has some advantages and some users adopted it for their own blogs (https://plopezadeva.com). If you strive for minimialism (with a touch of style) I can see hardly how you can more minimal. If you do not like the colors you can always change them. The typeface and the colors are defined in CSS, so you can provide those of your likings. For more fancy workflows a user developed a Markdown converted so that he can write his articles in TeXmacs and then use them with Hugo (here the result https://paperwhy.8027.org)

The graphics editor was buggy, but I find quite stable now. Lately it never blow in my face and if a speaker is writing on a blackboard I can essentially copy his drawings. If I'm lecturing on Zoom I usually use the graphics editor to illustrate some ideas ( see some examples here https://www.iam.uni-bonn.de/fileadmin/user_upload/gubinelli/... , note that this is the actual script of the lecture, written online and not postprocessed )

I'm not sure I understand a use case for lossless reverse HTML import. TeXmacs can import HTML (e.g. Wikipedia pages) including proper recovery of mathematical formulas from MathJaX or MathML. TeXmacs can export HTML with math in MathML (since 2005), MathJaX or images according to user preferences. The exported HTML human-readable (although with some CSS here and there) like the LaTeX export (who say the contrary has visibly never used TeXmacs, at least recently). You can check the quality of the HTML export by looking at TeXmacs website (https://www.texmacs.org/tmweb/home/welcome.en.html) and at the blog (https://texmacs.github.io/notes/docs/main.html) or also at this other blog (https://plopezadeva.com) or at the webpage of Joris van der Hoeven, the main developer (http://www.texmacs.org/joris/main/joris.html). Comments and suggestions for improvements are welcome. TeXmacs do not use UTF-8 (or any other Unicode encoding) internally but support an arbitrary, extensible encoding (much like Unicode) which can be converted to Unicode without problems. The reason is simple: TeXmacs was initiated before Unicode was widely supported (~1998). Even in current times we see few reason to move to Unicode, all the TeXmacs internals are independent from external libraries or tools. It was properly working in ~1999, it works now and it will work as long as a minimal C++ compiler works.

Thanks. Glad you liked it enough to remember :) Indeed, I've just checked and I still have the TeXmacs files for the "polycopié du cours de Processus Discrets". The main developer was in Orsay at the same time I was there 2006, for mixed circumstances I've started to interest me to that software at the same time. I should say at the beginning I was attracted to TeXmacs not as a user but because it is a very complex piece of software which is very fun to hack on (choose your language: Scheme or C++).

I convince them to use TeXmacs :). At least I do not want to use LaTeX anymore (note that I've been hardcore LaTeX users for long time and also developed software to support LaTeX, e.g. TeXniscope, a previewer for DVI/PDF files in the early times of MacOS). Conversion to LaTeX format happens in the last stage of the submission (to arXiv or to a journal). It takes usually from 1 to maybe 10 min to have the file ready, maybe small tweaks for some bugs which I report and then get corrected. Three of my students wrote their PhD thesis with TeXmacs, and at least two/three other people I know. If I need to collaborate on a file which is being written in LaTeX I write my parts in TeXmacs and then I paste them in the file.

Most of my presentations are also written in TeXmacs and in these days I use it to teach via Zoom, or to discuss with colleagues/students, make computations, etc... Not really using paper anymore. If you need to do algebraic manipulations, then it is more convenient to copy/paste on the screen that to rewrite the same equation several times with small variations on the paper.

Using the submission requirements as an argument seems so wrong to me, it is like saying: since the final format is PDF we need to write our ideas directly in PDF... if one day editors will decide to use XML as submission format, then what? I do not want to think about mathematics with a screen divided in two parts, and the quality of the process is also very important to me. I spend maybe months to work on a paper in TeXmacs and at most only 1h in LaTeX when I really need to because editors are become used to require from the authors the work a professional typesetter would have done in the past.

TeXmacs has interfaces for many systems (https://twitter.com/gnu_texmacs/status/1382216780702175232) including R. Seems that currently R does not run, however I think a medium R user can understand and fix the problem. The interface code for R is written in R.

I generally use TeXmacs for notes, derivations, lectures, articles, all, of course. Most of my papers in the last 7 years, around 15 and all my lectures in the same period, circa 2 courses a year.

More in detail, all these papers are been written completely in TeXmacs and then converted to LaTeX to be put in arXiv: https://arxiv.org/abs/2004.09637 https://arxiv.org/abs/2004.01513 https://arxiv.org/abs/2004.00872 https://arxiv.org/abs/2003.14264 https://arxiv.org/abs/1912.06881 https://arxiv.org/abs/1912.04830 https://arxiv.org/abs/1906.11187 https://arxiv.org/abs/1812.04422 https://arxiv.org/abs/1810.12014 https://arxiv.org/abs/1805.10814 https://arxiv.org/abs/1708.03118 https://arxiv.org/abs/1702.03195 https://arxiv.org/abs/1701.07373 https://arxiv.org/abs/1610.07886 https://arxiv.org/abs/1508.03877 https://arxiv.org/abs/1502.00157

Similarly all the lecture notes in this page: https://www.iam.uni-bonn.de/abteilung-gubinelli/teaching

I think most of the discussion in this thread does not report any real current experience using it.

you can opt for MathML or Mathjax output instead of images, but the majority of the math typesetting is more faithfully exported via images.

I didn't knew MPS but indeed seems quite similar. Both are structured editors, in the sense that you never work with "text" but with some projection of the underlying tree structure. For example you can ask TeXmacs to show you a representation where every node tree is explicit (which look very much HTML, but where you cannot have an open environment tag without the corresponding closing tag).