Knitr + pandoc Workflow for Executing Code and Insert Output (R Markdown)

[LucunJi]

Nice workflow, at least for me

> Code Here <

An competitive alternative for running codes before current extensions (Jupyter/Execute Code) become more mature.

This workflow has 3 parts: one to knit the markdown, one to compile into PDF, and another for opening the PDF.

You can find recommended settings and descriptions for custom variables in that Gist.

Features

• Very configurable (set the custom variables, configure the defaults file for pandoc, or just edit the code)
• Multiple languages (r, python, as long as knitr supports, but sorry there's no matlab)
• Per-code block settings for inputs, outputs, engines... (see document of knitr)
• Inserts images and outputs for saving & exporting
• Also supports syntax ```language +x{args} because Obsidian does not highlight R Markdown syntax ```{language args}
• Combining with QuickAdd, we can save file before kintting/exporting

Possible improvements

• Easier re-running code block: preserve per-block settings, clear the outputs
• Quickly making a code block executable
• Lua pandoc filters for centering images, scaling images in Obsidian's style (![caption|size](url))
• Auto-detect attachment path from .obsidian/app.json
• Make an extension for this, as the current code is already growing large

[LucunJi]

Note: run

mkpipe .Rpipe;
Rscript -e "while (T) {tryCatch(eval(parse(text=readLines('stdin'))), error = function (e) {print(e); sys.calls()})}" < .Rpipe &

Makes it possible to use variables across runs bu reusing the same R session with a FIFO.
It's a possible modification to the existing shellcommand.

[Taitava]

Nice! 🙂 I'll reply just shortly now. I need to look into this at a better time. Can you elaborate a bit about what is knit? I'm not familiar with R programming language, either, although I've heard the name.

Do you have some practical example use case on which you use this? 🙂 I'm just trying to familiarize myself with this workflow and its possibilities. 🙂

[LucunJi]

• RMarkdown as a format: just a dialect of markdown, usually has extension .Rmd rather than .md, allowing to mark code blocks that we want to run.
• RMarkdown as a system: a bunch of things for .Rmd file, including format specification, runing code, and exporting to PDF.
• RMarkdown as a R package: build the .Rmd file into a PDF via pandoc and knitr, also supports other formats.
• knitr: an R package for running R code in .Rmd and generating a .md file with output. You can use it by just calling knitr::knit('filepath'). It can do more things such as run code in LaTeX and run Python code.

I actually prefer Python/Matlab over R. But for running some simple code in markdown, RMarkdown seems to be competitive.