datar not working on RStudio notebooks

[rleyvasal]

I am trying to run code on RStudio R notebooks with datar but the code does not run. I would like to use R notebooks to highlight parts of the datar code and run separately.

@pwwang what is the issue with running datar on RStudio.

[pwwang]

This is because the source code is not available for datar/pipda to detect the piping syntax.
See also the following for solutions:

https://pwwang.github.io/datar/caveats/blind/

[rleyvasal]

Hi @pwwang, I did look at the documentation and did figure it out yet. can you provide an example to do the following in RStudio notebook ?

```{python}
from datar.all import * 
from datar.datasets import mtcars
mtcars >> mutate(carb_2 = f.carb)
```

[pwwang]

Here you go:

```{python}
from pipda import options
from datar.all import *
from datar.datasets import mtcars
options.assume_all_piping = True

mtcars >> mutate(carb_2 = f.carb)
```

[rleyvasal]

It did not work, look at the snapshot below:

[pwwang]

Ah, my bad. options.assume_all_piping = True should go before the datar importing.

The following should work:

```{python}
from pipda import options
options.assume_all_piping = True
from datar.all import * 
from datar.datasets import mtcars

mtcars >> mutate(carb_2 = f.carb)
```

[rleyvasal]

It worked, thanks @pwwang

[antonio-yu]

@rleyvasal @pwwang

If in Rsutio with the latest version , it both works whether options.assume_all_piping = True goes before the datar importing or not.

[antonio-yu]

@pwwang @rleyvasal
In the latest Rstudio (version 2021.09.0 ),it both works whether options.assume_all_piping = True goes before the datar importing or not.

{python}
from datar.all import  * 
from pipda import options,register_verb
from datar.datasets import mtcars
options.assume_all_piping = True 
mtcars >> mutate(carb_2=f.carb)

[pwwang]

@antonio-yu
Good to know. I believe it's more related to R/reticulate version.
Theoretically, it should work no matter options.assume_all_piping = True goes before or after datar importing.

[rleyvasal]

@antonio-yu, I can confirm that in the latest version of Rstudio (version 2021.09.0 ), options.assume_all_piping = True works irrespective of where it is entered - this is not an issue anymore.

[pwwang]

Then I will close this for now. If you have any related questions, feel free to reopen it or open new issues.

[rleyvasal]

@pwwang, just curious, do you know why the pipes work without pipda configuration in VSCode's Jupyter notebooks? but in RStudio notebooks we have to enter the following:

```{python}
from pipda import options
options.assume_all_piping = True
```

[pwwang]

It's all that matters whether the environment (jupyter, rstudio, etc) exposes the source code at runtime. datar/pipda depends on executing to detect the node to tell whether the verbs are calling regularly (e.g. dim(df)) or by piping (df >> dim()). executing did not work for jupyter before, but later it adopted jupyter environment. I will see what I can do with the Rstudio environment.

[pwwang]

@rleyvasal Just took a look at how Rstudio is running python. It actually executes python code with something like eval() wrapped by reticulate. It does not ship with the source code. So currently we need that assumption to let datar work. But I will keep on an eye on reticulate's evolution, and see if it'll be possible in the future. And maybe submit an issue to the repo at some time point.