How to use markdown in R

[DeadlySquad13]

Hi, thanks for your plugin, enjoying it a lot!
I have a question regarding using markdown with R kernel. In R there's no block comments so recommended format from python """%% throws error.

Well, it's not so convenient but I tried using line comments format like in jupyter_ascending:

# %% [markdown]
# # Heading
# Simple Text

But it got rendered as:

I have found a workaround specifying nolint: error:

"""%% # nolint: error.
Testing out markdown
%%"""

but it disables errors for all file.

Please help me how to do use it in R? Maybe I'm missing some option.

[kiyoon]

Thank you. Since I don't use R I can't verify this, but can you try:

# %% [markdown]
# nolint start
# Heading
Simple Text
# nolint end

Found from REditorSupport/languageserver#282 (comment)

[kiyoon]

Thanks for trying it out. I think the only good solution would be to ignore leading # in markdown cells on the plugin side, just like jupyter_ascending (jupytext).

[DeadlySquad13]

Yes, it seems like a best way. I would highly appreciate if you implement it. If you're short on time, please provide some helping info on how can I do it myself.

I have also found rmd format. Both lsp and styler in particular have support for it and I have tested that they work fine with it. Unfortunately, it's not compatible with notebook format though cells are delimited by triple backticks so I suppose it's possible to implement a custom parser for it. But I still think that implementing workaround for plain R files is better as I don't trust nvim's treesitter performance for injected languages. Even though it was improved yesterday.

[kiyoon]

@DeadlySquad13 Done in #53! Check it out :)

[DeadlySquad13]

Thanks, it was fast! I will test it on week.

By the way, I've been digging through your code trying to implement rmd support. Even though I said it might not be the best choice performance wise, it still piqued my interest. However, I had struggles with detecting first node as it's just plain markdown without special cell indicators. I added condition to add first markdown node if the first code cell found was of rmd format (```{r}). But because this class works with partial updates, it was spamming useless markdown blocks before each new code cell. With your changes from #53 I think it now can be fixed? As I see, we can differentiate between partial buffer analyse and full buffer analyse?

[kiyoon]

It could differentiate partial update and full update already, but the difference is that, previously it didn't need to know which cell it belongs to for the partial update. After the update we can now let it know which cell type it belongs to. So if the cell type is Jupytext markdown I just removed all leading comments.

Current ju.py format parsing is like, before the first separator it is ignored (header cell). It doesn't really detect the filetype and it is an universal format for all languages.

For rmd file, I think we need to detect ```{r} and if that exists we need to set the first cell type to markdown.

[DeadlySquad13]

Everything works fine so far. Thanks, @kiyoon for quick update of plugin!

---

Wouldn't mark first comment as answer to avoid confusion.