I'm also building a blog with Typst, but I don't use the html export or the typst binary at all.
Instead I only use it as a markup language and use the typst libraries to evaluate it into blocks/AST which I then use together with askama templates.
This gives me a lot more flexibility, and typst is nicely extensible with custom typesafe functions. The downside is that existing LSP integrations don't recognize my custom keywords and show errors, something I just accept for now but might work around by importing a lib.ty stub (without real implementation, just matching signatures).
Cool idea. So you essentially only use the Typst parser (i.e. the generated AST, I assume the text format of it)? With askama templates you essentially implement your own Typst renderer, which sounds like quite an achievement. May I assume that only a subset of Typst is supported by your renderer? What is the advantage of all this effort compared to e.g. Markdown or Asciidoc?
The HTML target being unstable, meaning I can't really build a half-decent ePub target from it, is probably the key blocker for me strongly considering Typst for my workflow.
For now, it's an unhappy marriage of pandoc, LaTeX, and a 20k LOC Lua ecosystem for me.
I'm also building a blog with Typst, but I don't use the html export or the typst binary at all.
Instead I only use it as a markup language and use the typst libraries to evaluate it into blocks/AST which I then use together with askama templates.
This gives me a lot more flexibility, and typst is nicely extensible with custom typesafe functions. The downside is that existing LSP integrations don't recognize my custom keywords and show errors, something I just accept for now but might work around by importing a lib.ty stub (without real implementation, just matching signatures).
Cool idea. So you essentially only use the Typst parser (i.e. the generated AST, I assume the text format of it)? With askama templates you essentially implement your own Typst renderer, which sounds like quite an achievement. May I assume that only a subset of Typst is supported by your renderer? What is the advantage of all this effort compared to e.g. Markdown or Asciidoc?
The HTML target being unstable, meaning I can't really build a half-decent ePub target from it, is probably the key blocker for me strongly considering Typst for my workflow.
For now, it's an unhappy marriage of pandoc, LaTeX, and a 20k LOC Lua ecosystem for me.