This package is pinned to versions of pandoc-types (>=1.17.6.1 && <1.18 || ==1.20.*) while Hakyll 4.13.4.1 now requires pandoc == 2.10.* which in turn requires pandoc-types == 1.21.*.

I tried compiling with the newer pandoc-types anyways using Stack's allow-newer: true option, and everything still seemed to work fine. I also tested with pandoc-types == 1.22.* and that worked as well. Can we bump dependency version for this package to be pandoc-types (>=1.17.6.1 && <1.18 || >=1.20 ) perhaps?

In latex-svg-image, adding this modifier to the document definition as follows:

\documentclass[12pt,dvisvgm]{...}

Allows tikz pictures to be later on rendered by dvisvgm.

Would you consider adding it?

For reference: this is where I found about it https://tex.stackexchange.com/a/560548

https://groups.google.com/forum/#!topic/hakyll/vQpijNtTG5M

Hello Oleg,

Thank you for making this fork. I don't know if you have noticed, Pandoc can produce RawInline and RawBlock elements for plain TeX strings inside a markdown document.

For example, the two LaTeX environments in

Inline environment,
\begin{align*}
  a &= 1 + 1\\
    &= 2
\end{align*}
rest of the paragraph.

\begin{tikzpicture}[>=stealth]
  \draw[->] (-1,0) -- (1,0);
  \draw[->] (0,-1) -- (0,1);
\end{tikzpicture}

will be compiled to the following Pandoc ASTs, respectively,

RawInline (Format "tex") "\\begin{align*}\n  a &= 1 + 1\\\\\n    &= 2\n\\end{align*}"
RawBlock (Format "tex") "\\begin{tikzpicture}[>=stealth]\n  \\draw[->] (-1,0) -- (1,0);\n  \\draw[->] (0,-1) -- (0,1);\n\\end{tikzpicture}"

which means it's possible to generate images for any TeX strings inside a markdown document with two filters of type

Inline -> Inline
Block -> Block

and put the raw TeX string into here.

By the way, the Pandoc AST can be generated using

$ pandoc -t native test.md

I was looking at the code to see if it's possible to extend it to support general LaTeX environments, but it seems that some major refactoring is needed because currently environments are tied to MathType. Also, "formula" is in most functions' names :). It probably doesn't make sense to make another fork of this. Could you consider implementing this feature? I might be able to help but there can be some major design decisions to be made.

Best,

Specifically alterForHTML should strip that id.