Typst 0.14: Now accessible
104 points by skrzyp
104 points by skrzyp
I finally used Typst for my first small project, recently, after following it for a long time. It was exactly what I hoped, and actually surpassed my expectations: I did the entire thing over a weekend, on my phone, on the couch rather than hidden away in my office. The web interface worked great!
Character-level justification looks amazing. The improved HTML export has me considering a few more projects. Latex scared me too much to ever even try.
I love the Markdown-ish with programming syntax of Typst and have, after very little usage, already reached further than I ever dared to go with LaTeX.
It's a really solid and well-documented tool.
Looking forward to the HTML export!
I left academia just before typst started seeing any serious use, and so I have the pleasure of being able to use typst in personal aspects.
I remember being annoyed by a few small things, but I can't remember what. For my usage, this is a worthy and welcome successor to LaTeX.
I really like Typst and I've already used it for my bachelor thesis and I have no regrets about not using LaTeX. The language is nice and practical and very well thought.
I only have a few minor issues with some missing features. The last time I checked figure floating and general wrapping around objects is missing (even if there is a package called wrap-it that kind of solves it user-land) and I still can't use the "inkscape .pdf_tex workflow" for creating pictures and diagrams with plain text and math inside rendered by typst directly.
Really the only thing Typst is missing is first class support for custom elements, though there has been some progress on that in the background. Once that's in place, I think it'll really be both more powerful and more ergonomic than Latex.
By custom elements do you mean on-the-fly (or typst-package) based syntax changes to the markup?
Typst lets you apply styles through set rules and show rules. These can only be applied to certain blessed built-in functions. Typst calls these functions "elements".
There was a proposal a while ago for letting users define their own elements that would have first class styling support. When this feature is added, it'll likely become the de facto way many packages expose their APIs.
I'm intrigued but not sold on it yet. I've been using LaTeX since the 1990's. On the one hand want something easier, less obtuse, and faster to run and on the other already have sunk a lot of time and tears into what I have and things like tectonic and LuaTeX work nicely.
You should know that Typst already has a vibrant package ecosystem, and supports sandboxed compiler plugins in the form of WebAssembly.
Typst also already does many things functionality-wise decidedly better than LaTeX, not just convenience and speed.
Having sunk time into LaTeX shouldn't prevent you from trying Typst. You can later go back to LaTeX and your blood, sweat, and tears will still be there, if you want to.
You should know that Typst already has a vibrant package ecosystem.
As a Typst (and LaTeX) user myself, I think that this is an exaggeration. I often encounter situations where there is no package (or primitive support) for doing what I need, and sometimes the one package available is primitive compared to the LaTeX counterpart I am used to and has clear issues. I also observe that the Typst language and programming conventions are currently not so good at producing easy-to-customize packages, so for some domains there are often many different packages or templates that are suited to their authors' preference, but realistically users need to redo their own.
(For example, thesis templates: every student who decided to write their master thesis in Typst apparently create a new template for their own university, whereas you would expect a main "thesis" template that can be customized with extra options, at least as a common basis between all of those.)
I have discussed this on the Typst forum on several occasions (for example: Overriding template parameters: missing social convention or typst design flaw?), and I am afraid that there is no short easy fix, this is an ecosystem maturity issue that needs time to sort itself out -- and maybe the language needs to grow extra features.
I think it’s funny that in that thread is proposed some sort of style sheet that cascades
I have little experience with it but my first impressions of Typst were great! Before its existence I always dreamed of a better, more thought out alternative to LaTeX. The only thing I find mildly annoying about Typst is not having the same math mode syntax as LaTeX. I wish I there was a way to enable a math-compatibility mode.
It's a pretty wild approach to the problem, but you might be interested in MiTeX! https://github.com/mitex-rs/mitex
How is typst doing on support for mathematical notation? And does it yet make such notation screen-reader-accessible in any way?
When I played with typst its support for math notation seemed sufficient for everything I've ever done with math notation in latex.
The accessibility guide says
Finally, you can specify an alternative description on math using math.equation. Describe your formula as if read out loud in natural language. Currently, adding an alternative description is required for accessible math for all export formats. Not adding an alternative description for your formula will result in a failure of PDF/UA-1 export. In the future, Typst will automatically make math accessible in HTML and PDF 2.0 by leveraging MathML technology.
I.e. right now you can apply alt text to math equations, and they hope to do better in the future.
If anyone is interested in an example of a Typst book: https://github.com/bigskysoftware/hypermedia-systems-book