Debug in R markdown file

[taooceros]

Is it possible to debug for a code chunk within a r markdown file? I think python jupyter notebook does support the feature, same as r studio.

[ManuelHentschel]

Unfortunately, this is not possible at the moment. Implementing this properly is going to be difficult (e.g. due to all the different settings/commands that can be specified for each code chunk).

To still allow some Rmd debugging, I can think of the following two alternatives:

• Start a debug session with an empty workspace and send the code inside a chunk to the debug console (e.g. using ctrl+enter or the buttons provided by the vscode-R extension). To an extent this might already work.
• Convert the entire Rmd file to R using e.g. knitr::purl() and debug that. This would ignore the .Rmd specific settings and division into chunks, but allow you to run the R code itself in the debugger. It might also be possible to implement this, so that the conversion is done automatically in the background.

[taooceros]

Start a debug session with an empty workspace and send the code inside a chunk to the debug console (e.g. using ctrl+enter or the buttons provided by the vscode-R extension). To an extent this might already work.

That looks reasonable, and I remember the jupyter notebook do the same thing.
Convert the whole document to r will lose the advantage of rmd. Though I think the second option may be easier.

[markusgumbel]

I wonder if you could consider a chunk just a piece of code to be executed in the global environment. What variables are defined at debugging time is up to the user. He/she could run any chunks above (that would create those global variables) before debugging whenever needed.

Debugging a chunk is a very important feature in my opinion. Jetbrains' R plug in can do so - RStudio not.

But anyway, VS code's R features incl. this debugger are already excellent! Please go on!

[JZL]

I agree and for what it's worth, where I find using the browser() debugger useful in rstudio .rmd files is to debug external libraries/other R files, stepping through the function calls to see where an error/bug is coming from. So I wouldn't need/want anything very crazy (or even cross rmd cell boundaries) -- just pulling out the current cell into an execution environment along with all the global/local vars.

If I made the plunge to use VScode over Rstudio I could definitely use purling to get it to work but I wanted to be careful about not losing the current environment.

[milanglacier]

Start a debug session with an empty workspace and send the code inside a chunk to the debug console (e.g. using ctrl+enter or the buttons provided by the vscode-R extension). To an extent this might already work.

That looks reasonable, and I remember the jupyter notebook do the same thing. Convert the whole document to r will lose the advantage of rmd. Though I think the second option may be easier.

purl() exported .R file will have your code chunk separated by ## -----------------,

and you just simple replace ## ----------------- by # %%, and you can use the command palette run current chunk , which makes a R file runs blockwisely like rmd