Typst: a Possible Latex Replacement

Almost nobody wants to learn something new when they already know something similar.

Creates a heck of a momentum effect, not just from the practitioners resisting the change, but also available resources and so on.

1. It doesn't generate 5 bloody files when compiling.

2. Compiling is instant.

3. Diagnostics are way easier to understand (sort of like Rust compiler suggestion style).

4. List items can be either - item1 - item2, etc. or [item1], [item2]. The latter is way better because you can use anchoring to match on the braces (like "%" in vim), which means navigating long item entries is much easier.

5. In latex you have the \document{...} where you can't specify macros so they need to be at the top, in Typst you can specify the macros close to where you need them.

6. It's easier to version control and diff, especially if you use semantic line breaks.

7. Changing page layout, margins, spacing between things, etc., footers with page counters, etc. just seems way easier to do.

HTML with MathML may make a decent system as well; possibly with an XML source and XSLT for templating, which is apparently how OpenStax textbooks are composed (via CNXML, though that also has just a couple of rain frog pictures in its documentation repository -- seems to be a common pattern around typesetting systems).

Then there is troff with eqn(1), which looks simpler, but not sure if there is an actual specification for it around, either.

And then there are Texinfo, org-mode with LaTeX embedding, other TeX-adjacent options, perhaps Markdown with HTML and MathML embedding.

If I want to write C# on my resume, I do C`#` but there has to be a better way?

Pros:

- Instant compile. It just sits there waiting, and once you save your .typ, boom, your .pdf is ready.

- Surprisingly often I find myself using it as markdown replacement, e.g., for random meeting notes. Syntax is as easy as markdown and without boilerplate it produces a nice pdf.

What's not to like?

- IMO debugging can be tricky with quite concise error messages. And it does not produce any pdf once there is a single syntax error, precluding one favourite latex debugging route.

- When using packages, one does encounter hickups, but no surprise here for long-time latex users.

So asking the community here: what does Typst offer in place of PGF/TikZ[1], PGFPlots[2], Asymptote[3], chemfig[4], siunitx[5], physics2[6], and how does it work with existing bibliography providers? I use biber[7] with the Zotero Connector and Better BibTeX[8] so any paper I visit on the web is essentially instantly available to cite with one click on LaTeX.

A good test for Typst ought to be reproducing most of these typographic and diagrammatic exemplars: https://tex.stackexchange.com/questions/1319/

[1]: https://tikz.dev/

[2]: https://tikz.dev/pgfplots/

[3]: https://asymptote.sourceforge.io/

[4]: https://mirrors.ctan.org/macros/generic/chemfig/chemfig-en.p...

[5]: https://mirrors.ctan.org/macros/latex/contrib/siunitx/siunit...

[6]: https://mirrors.ctan.org/macros/latex/contrib/physics2/physi...

[7]: https://mirrors.ctan.org/biblio/biber/base/documentation/bib...

[8]: https://retorque.re/zotero-better-bibtex/

It would be nice if Typst could be a LaTeX replacement that makes actually layout and designing the document approachable. I have only used it once for a quick one-off experiment and I did like the language, but as I have said above the language is not the problem if you just want to write text.

[1] That's not quite true, actually. I first wrote my thesis in reStructuredText and used Pandoc to generate the LaTeX and subsequent PDF. This allowed me to get started without having to write a lengthy preamble first. Then after I had more than half of it written down and had a good idea of what I wanted the document to look like did I clean up the generated LaTeX, adjust the formatting to my needs, redid the drawing in TikZ, and then kept writing LaTeX from there. I still think the language is not the problem, but it's easy to get hung up in the design phase before even the first chapter is written.

The output is superb though, which is why everyone in academia keep using it. It’s just the tooling that is poor.

I think it needs to be modernized.

I then discovered typst and it is a breath of fresh air. Unicode/Dēvanāgarī support out-of-the-box, no installing gigabytes of packages, near-instant compilation.

My complements to those who got this done.

* Submitted too fast. A quick search tells me Pandoc already added Typst input and output support (e.g. https://pandoc.org/typst-property-output.html), so guess I need to look into if I should switch to use that for generating PDFs.

[0]: https://ezb.io/thoughts/interaction_nets/lambda_calculus/202...

We should just use human friendly markups like MyST Markdown [1] or Org Mode [2].

Unfortunately, whitepapers are predominantly written in LaTeX. Thankfully, arXiv recently made a move to parse and render those documents in the web format. It's a hard job. But this is the wrong way around: instead of keep composing documents in LaTeX (which is not human friendly), and then doing the hard job with tooling, we should start with human first approach and have win-win!

We are living in the world where web content is the primary content and friendly for desktop, mobile devices and readers and tools (select, copy, edit...). It's easy to package any web content into epub and ship it in a single file. Printing is also easy. Only cons: precise typesetting is not harder. But this is less of the problem. I would prefer a content that is friendly to read and is responsive, than a precise typesetting.

[1] https://mystmd.org/ [2] https://orgmode.org/manual/Summary.html [3] https://info.arxiv.org/about/accessible_HTML.html

Does it make the blogs look more edgy and geeky? Is it a an artistic trend in web design?

I’d love some insights from someone with a better sense of the situation

I am also worried about the rust-centricity, seeing how rust is somewhat of a moving target.

Our previous pipeline was LaTeX-based (first pdflatex, then lualatex), but we were constantly fighting cryptic memory errors on large documents and huge Docker image sizes that slowed down boot times of our ephemeral workers.

Switching to Typst was a massive win for us. The single static binary resulted in tiny images and faster boot times. More importantly, the performance gains were huge. Overall compile times were ~3–4× faster than LaTeX. On really large documents (2000+ pages, mostly tables), Typst compiles in ~1 minute vs. ~18 minutes with lualatex.

Beyond performance, the better developer experience and good error messages was a nice bonus too.

We wrote a detailed post about the entire architecture - from the job orchestration with Nomad to the S3 optimizations and the Typst migration in particular. If you're curious, you can read it here: https://zerodha.tech/blog/1-5-million-pdfs-in-25-minutes

Not having to deal with the insanity of the LaTeX distribution system alone is worth the switch. Everything is contained in a single binary. The language itself is much simpler to read and write, and seems just as flexible. LLMs do a decent job of generating it. Compilation takes less than a second, making it so much faster to iterate.

Many thanks to the authors and contributors, and please don't ruin it. :)

But I will say I've mostly written relatively simple documents in it, so maybe that colors my experience.

[1]: https://github.com/scalawithcats/scala-with-cats/tree/featur...

The absolute staying power of TeX has been incredible.