What's the differences between using a visual pkg and inserting images generated by a (external) visual pkg?

Questions
, ,
1

Perhaps you know:

  • There are some classic visual packages, such as Matplotlib.py, PlantUML.jar, D3.js, GR.c.
  • In LaTeX CTAN and Typst Universe, there are TikZ, PGFPlots, CeTZ and so forth.

So, what is the most key difference between them?
In LaTeX, I see people plot in this way:

\usepackage{python}
import matplotlib as mpl
plot = mpl.plot(blah blah)
save plot as pdf
insert pdf as image into figure
\end{python}

Does it make sense (in Typst)?

2

Essentially it’s a matter of preference. If it can exist, it eventually comes to existence. LaTeX already has so many features bundled as packages, and I suppose TikZ just so happened to land in the same category for Typst, under the CeTZ branding. Though Typst does also come with an underrated curve function.

Most notably for TikZ/CeTZ, these features are argued about:

  • Users don’t like to swap between applications for drawing like Inkscape and writing.
  • Less files, because the images don’t have to be exported.
  • Requirement to submit only .tex/.typ.
  • Fonts in the image exactly match the surroundings, which can be more difficult to achieve with external tools like Inkscape.
    • This is especially true for plotting, where you likely want matching line thickness and font size. Not to mention the adjustable grid, ticks, background colour, and so on.

Being its own programming language, Asymptote is both a package and an application that offers powerful plotting for LaTeX. So even if the images generated by Asymptote are imported from outside, fonts can easily be matched. There’s nothing like it in Typst as of now, and I don’t think we’ll see that any time soon, the closest in terms of billboard text would probably be the Typst addon in Blender.

Another option that I know of for Typst would be importing an image as SVG or PDF, then replacing its parts in the code:

Turning the attention more to the essence of your post, I believe Typst already can run many programming languages, even LaTeX. Mostly through packages which introduces many limitations. So I’m not certain about the plotting aspect in that way, but for that, Typst’s lilaq package is aiming to become natively compatible with Python’s matplotlib if I remember correctly.

At least in Typst, all these options are still very much evolving.

2 Likes