I can't attach any R session to VS Code when using a remote server via SSH and it can't find the binary either.

[rgriva]

I am using a conda virtual environment for a research project conducted at a remote server which I access via SSH.

I use the [Remote SSH] (https://marketplace.visualstudio.com/items?itemName=ms-vscode-remote.remote-ssh) extension to connect to the host and then I need to edit R files there. The code below shows that both R and radian are correctly set, I believe:

(r_conda) [rgr6291@klc0201 bin]$ conda env list
# conda environments:
#
fomc                     /home/rgr6291/.conda/envs/fomc
r_conda               *  /home/rgr6291/.conda/envs/r_conda
rfe_paper                /home/rgr6291/.conda/envs/rfe_paper
base                     /software/python-anaconda3/2019.10

(r_conda) [rgr6291@klc0201 bin]$ which R
~/.conda/envs/r_conda/bin/R
(r_conda) [rgr6291@klc0201 bin]$ which radian
~/.conda/envs/r_conda/bin/radian

In the settings for the R extension, I have set these binaries as the RPath and the Rterm:

{
    "r.rpath.linux": "~/.conda/envs/r_conda/bin/R",
    "r.rterm.linux": "~/.conda/envs/r_conda/bin/radian"
}

However, I get errors like this one when I try to attach an R session:

Cannot find R client at ~/.conda/envs/r_conda/bin/radian. Please check r.rterm setting.

And also it complains that it can't find the jsonlite package although it has been installed:

Failed to get list of R functions. Make sure that jsonlite is installed and r.rpath.linux points to a valid R executable.

What am I missing here?

[albertosantini]

I have not the same setup, but I would check if vscode-r is remotely installed.

See also https://code.visualstudio.com/docs/remote/ssh#_managing-extensions

The second guess is using an absolute path.

Then I would check the extension output to see if there is something to exploit.

Hope that helps.

[rgriva]

I do see a .vscode-r folder on my home folder. How could I check everything is correctly installed?

And Googling other options I just realized I don't seem to have an .Rprofile file. How would that impact these errors?

[albertosantini]

In vscode extension panel (ctrl+shift+x), but if you see that folder it is likely to think the extensions are remotely installed. I suppose you intend, for your home folder, remote home folder.

Please, check again from vscode and go to the second guess (absolute path). Don't forget to give a look at the output panel (in vscode).

.Rprofile is not a term of the equation.

[rgriva]

So, I tried absolute paths and nothing changed.

When I try to attach to the R session, for example, it gives me the error

Error in .vsc.attach() : could not find function ".vsc.attach"

Indeed, if I run the following script line by line nothing will show up on the screen, although I get completion suggestions:

library(tidyverse)
library(ggplot2)

df <- mtcars
df %>% ggplot(aes(x = mpg, y = wt)) + geom_line()

Also, I checked on the remote and it seems to be correctly installed:

Any ideas on what might be going on?

[renkun-ken]

For conda environments, the shell profile should be sourced to inherit necessary environment variables.

Maybe related: https://github.com/REditorSupport/vscode-R/issues/431.

[albertosantini]

I tried that extension (remote-ssh) against a linux machine with R installed and vscode-r works fine. In my case R executable is in the path (no RHOME or vscode settings set).

In your case vscode-r extension is installed and enabled correctly on remote machine.

I would temporary forget the conda setup and I would give a try with R in the path. I don't know conda and I don't know if there is a special meaning for (r_conda) in the prompt. If that is a standard linux prompt and which R returns a path (as you did), I don't understand why it doesn't work.

Basically is R installed in ~/.conda/envs/r_conda/?

[precedex28]

I am using anaconda as well not remote but on the PC and having the same issues I have tried absolute paths and it will not connect to the r.path or the radian

[albertosantini]

Ok. It is not related to remote ssh extension.

I don't know conda. How to reproduce your conda setup? Then I would try to reproduce the issue.

See also https://github.com/REditorSupport/vscode-R/issues/623, https://github.com/REditorSupport/vscode-R/issues/747. Any hint from these issues?

Did you ask any help in conda community?

[rgriva]

I tried today on my local machine. No conda environments involved. In fact, radian is by no means necessary for this extension so I avoided using it. So I avoided it completely.

My test setup today was the following:

• Fresh install of vanilla R.
• Fresh install of VS Code straight from Microsoft's website.
• Fresh install of the vscode-R extension.

Then, to set it up, I followed the instructions from this Youtube video: https://www.youtube.com/watch?v=EDJqHZx0JnQ&t=602&ab_channel=NickEubank

I noticed that, at least on my machine (M1 Macbook Air), the two following paths are different:

(base) ➜  ~ which R
/usr/local/bin/R
(base) ➜  ~ R  

R version 4.1.0 (2021-05-18) -- "Camp Pontanezen"
Copyright (C) 2021 The R Foundation for Statistical Computing
Platform: aarch64-apple-darwin20 (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

  Natural language support but running in an English locale

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

> R.home("bin")
[1] "/Library/Frameworks/R.framework/Resources/bin"

Namely, which R points to a different path than running R and then finding the path for R.home("bin"). The video I linked above says one should indicate the latter as the path for R.

After these steps, I get code completion and so on (as I did in the remote) and now when I plot something I get the respective plot on a side window inside VSCode itself, as it should be. Even now, if I click on "attach R session", I get the same error as before.

After doing this, I did install radian and tried to follow the steps in the video. I could not make VSCode work with that.

@albertosantini can you detail what steps you took? Can you attach to a current R session? What happens when you run, for example:

plot(mtcars$mpg, mtcars$hp)

[albertosantini]

Thanks for the details. On my side I have the following setup:

1. On the local machine vscode and R are installed. vscode-r extension works fine: I can open an R terminal or I can click on the statusbar to attach to active terminal.
2. On the remote machine I have R in the path.
3. Then I locally install remote ssh extension. I go to the ssh targets panel and open my remote machine. A workspace is opened and I can work like in local: opening a R terminal, attaching to active terminal without any setup on my side. If I execute that plot command, a graph is opened.

[eitsupi]

FYI, If you can use Docker, you can use the built-in R definitions in the Remote-Containers extension to see how vscode-R works, including radian. https://github.com/microsoft/vscode-dev-containers/tree/main/containers/r

[precedex28]

Thanks for the quick reply, could you try with Anaconda as it is a common (Free) platform used in data science for python and R. Majority of business data scientists are required to use it.

Thanks Scott

On Mon, Jan 24, 2022 at 8:05 PM Raul Guarini Riva @.***> wrote:

I tried today on my local machine. No conda environments involved. In fact, radian is by no means necessary for this extension so I avoided using it. So I avoided it completely.

My test setup today was the following:

• Fresh install of vanilla R.
• Fresh install of VS Code straight from Microsoft's website.
• Fresh install of the vscode-R extension.

Then, to set it up, I followed the instructions from this Youtube video: https://www.youtube.com/watch?v=EDJqHZx0JnQ&t=602&ab_channel=NickEubank

I noticed that, at least on my machine (M1 Macbook Air), the two following paths are different:

(base) ➜ ~ which R

/usr/local/bin/R

(base) ➜ ~ R

R version 4.1.0 (2021-05-18) -- "Camp Pontanezen"

Copyright (C) 2021 The R Foundation for Statistical Computing

Platform: aarch64-apple-darwin20 (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.

You are welcome to redistribute it under certain conditions.

Type 'license()' or 'licence()' for distribution details.

Natural language support but running in an English locale

R is a collaborative project with many contributors.

Type 'contributors()' for more information and

'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or

'help.start()' for an HTML browser interface to help.

Type 'q()' to quit R.

R.home("bin")

[1] "/Library/Frameworks/R.framework/Resources/bin"

Namely, which R points to a different path than running R and then finding the path for R.home("bin"). The video I linked above says one should indicate the latter as the path for R.

After these steps, I get code completion and so on (as I did in the remote) and now when I plot something I get the respective plot on a side window inside VSCode itself, as it should be. Even now, if I click on "attach R session", I get the same error as before.

After doing this, I did install radian and tried to follow the steps in the video. I could not make VSCode work with that.

@albertosantini https://github.com/albertosantini can you detail what steps you took? Can you attach to a current R session? What happens when you run, for example:

plot(mtcars$mpg, mtcars$hp)

— Reply to this email directly, view it on GitHub https://github.com/REditorSupport/vscode-R/issues/946#issuecomment-1020735329, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADM7ZKC5ANRIVGCP6MVPSSTUXYANVANCNFSM5MH6F5HA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you commented.Message ID: @.***>

-- Scott Taylor Pharm.D. MSc BCPS Informatics Pharmacist

[pgramme]

Hi !

My environment is the same as raulguarini (OP): VsCode + R extension + SSH remote shell (Linux) + R only available through a Conda environment on the remote. I haven't tried to use radiant. After struggling a bit I just found how to use the Conda env properly, so I'm sharing it here.

In my case, the Conda environment also contains R packages (tidyverse etc) and an ODBC driver that I need to use from R. In order to use syntax highlighting from VsCode, I installed r-languageserver and r-jsonlite in my Conda environment.

Three options: the first one is a bit easier but not complete enough for me. I went for Option 3.

I hope this can help someone else! If so, maybe document it somewhere? Pierre

Option 1

In the settings for the R extension, set RPath and the Rterm: (NB: I avoided "~" in the paths, not sure if expansion would work or not)

{
    "r.rpath.linux": "/home/myself/conda/envs/my_conda_env/bin/R",
    "r.rterm.linux": "/home/myself/conda/envs/my_conda_env/bin/R"
}

Result: R is available, tidyverse also OK but not the ODBC driver is not accessible from R.

Option 2

As suggested in #623, you can try to set "r.alwaysUseActiveTerminal": true in the VsCode settings. Then manually start a terminal, Conda-activate and start R. After that you should be able to submit R commands (e.g. with Ctrl+Enter). I find it less comfortable and didn't try it.

Option 3

Write a little shell script which first activates Conda and then starts R. I wrote it in /path/to/myproject/R-conda.sh and made it executable (chmod +x R-conda.sh)

#!/bin/bash
MY_CONDA_ENV=my_conda_env

# Optional and not related to Conda: cd to the script's dir
cd $(dirname $0)

# First we need to init conda bc ~/.bashrc is not executed when starting non-interactive shells. 
# Assuming PATH was already set by parent shell, and reusing script in profile.d/ if possible...
__conda_init_file=$(dirname $(which conda))/../etc/profile.d/conda.sh
if [ -f "$__conda_init_file" ]; then
    source "$__conda_init_file"
else
    eval "$(conda shell.bash hook 2> /dev/null)"
fi    

# Now we can activate
conda deactivate && conda activate $MY_CONDA_ENV

# And start R, passing through all command-line arguments
R "$@"

Then I pointed VsCode-R to this script in the settings:

{
    "r.rpath.linux": "/path/to/myproject/R-conda.sh",
    "r.rterm.linux": "/path/to/myproject/R-conda.sh"
}

Result: R is available, tidyverse and the ODBC driver are accessible from R.

[renkun-ken]

@pgramme Thanks for digging into this. Would you like to put your solution to https://github.com/REditorSupport/vscode-R/wiki/FAQ#how-can-i-work-with-r-provided-by-conda-environment-under-remote-development?

[pgramme]

@renkun-ken Sure! but maybe first have a confirmation that it also works for other people?

[renkun-ken]

Yes, could anyone confirm that it works?

[ashrosewood]

Hi !

My environment is the same as raulguarini (OP): VsCode + R extension + SSH remote shell (Linux) + R only available through a Conda environment on the remote. I haven't tried to use radiant. After struggling a bit I just found how to use the Conda env properly, so I'm sharing it here.

In my case, the Conda environment also contains R packages (tidyverse etc) and an ODBC driver that I need to use from R. In order to use syntax highlighting from VsCode, I installed r-languageserver and r-jsonlite in my Conda environment.

Three options: the first one is a bit easier but not complete enough for me. I went for Option 3.

I hope this can help someone else! If so, maybe document it somewhere? Pierre

Option 1

In the settings for the R extension, set RPath and the Rterm: (NB: I avoided "~" in the paths, not sure if expansion would work or not)

{
    "r.rpath.linux": "/home/myself/conda/envs/my_conda_env/bin/R",
    "r.rterm.linux": "/home/myself/conda/envs/my_conda_env/bin/R"
}

Result: R is available, tidyverse also OK but not the ODBC driver is not accessible from R.

Option 2

As suggested in #623, you can try to set "r.alwaysUseActiveTerminal": true in the VsCode settings. Then manually start a terminal, Conda-activate and start R. After that you should be able to submit R commands (e.g. with Ctrl+Enter). I find it less comfortable and didn't try it.

Option 3

Write a little shell script which first activates Conda and then starts R. I wrote it in /path/to/myproject/R-conda.sh and made it executable (chmod +x R-conda.sh)

#!/bin/bash
MY_CONDA_ENV=my_conda_env

# Optional and not related to Conda: cd to the script's dir
cd $(dirname $0)

# First we need to init conda bc ~/.bashrc is not executed when starting non-interactive shells. 
# Assuming PATH was already set by parent shell, and reusing script in profile.d/ if possible...
__conda_init_file=$(dirname $(which conda))/../etc/profile.d/conda.sh
if [ -f "$__conda_init_file" ]; then
    source "$__conda_init_file"
else
    eval "$(conda shell.bash hook 2> /dev/null)"
fi    

# Now we can activate
conda deactivate && conda activate $MY_CONDA_ENV

# And start R, passing through all command-line arguments
R "$@"

Then I pointed VsCode-R to this script in the settings:

{
    "r.rpath.linux": "/path/to/myproject/R-conda.sh",
    "r.rterm.linux": "/path/to/myproject/R-conda.sh"
}

Result: R is available, tidyverse and the ODBC driver are accessible from R.

Option 2 worked for me after conda installing r-languageserver and r-jsonlite - thanks so much for this post!! I like option 2 for constructing new code, but option 3 with command line args is probably better for debugging.

[jsta]

I'm still getting:

Failed to get list of R functions. Make sure that jsonlite is installed and r.rpath.linux points to a valid R executable.

Even with Options 1 + 2 in effect.

Edit

The issue was solved for me by upgrading my local R install to 4.2.0 and doing a fresh install of local packages (e.g. languageserver). 🤷🏽‍♀️

[Yunuuuu]

Just came across this problem, I also used conda to manage R in a remote server. Since the server has network limitations and I have no root permission, I can only work R with a conda, and set some local CRAN or Bioconductor mirror to install packages. here is the detail usage list:

• use vscode to connect the server
• install R with conda in the user path
• install package use R function install.package (since I don't want to use conda to manage R packages)
• use conda to manage the R package compiling dependencies
• use radian to provide Rterm

here is my sessionInfo() output:

R version 4.2.1 (2022-06-23)
Platform: x86_64-conda-linux-gnu (64-bit)
Running under: CentOS Linux 7 (Core)

Matrix products: default
BLAS/LAPACK: /home/user/anaconda3/envs/R-release/lib/libopenblasp-r0.3.21.so

locale:
 [1] LC_CTYPE=en_US.UTF-8      
 [2] LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8       
 [4] LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8   
 [6] LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8      
 [8] LC_NAME=C                 
 [9] LC_ADDRESS=C              
[10] LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8
[12] LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils    
[5] datasets  methods   base     

loaded via a namespace (and not attached):
[1] compiler_4.2.1   cli_3.4.1       
[3] tools_4.2.1      progressr_0.11.0
[5] jsonlite_1.8.3   digest_0.6.30   
[7] rlang_1.0.6     

I install R by command conda create --name R-release -c conda-forge r-base git radian here is my setting:

• vscode setting (don't use "~", it seems vscode won't expand it, just use a absolute path)

    "r.alwaysUseActiveTerminal": true,
    "r.bracketedPaste": true,
    "r.lsp.debug": false,
    "r.rpath.linux": "/home/user/anaconda3/envs/R-release/bin/R",
    "r.rterm.linux": "/home/user/anaconda3/envs/R-release/bin/radian",
    "r.rterm.option": [
        "--r-binary=/home/user/anaconda3/envs/R-release/bin/R",
        "--local-history",
        "--no-save",
        "--no-restore"
    ]

• .Rprofile setting

# local CRAN and Bioconductor mirror
local({
    # set a CRAN mirror
    r <- getOption("repos")
    r["CRAN"] <- "https://mirrors.tuna.tsinghua.edu.cn/CRAN/"
    options(repos = r)
    options(BioC_mirror = "https://mirrors.tuna.tsinghua.edu.cn/bioconductor")
    # since the network limitation of my remote server, I used a local `bioconductor_config.yaml`
    # downloaded from https://bioconductor.org/config.yaml
    options(BIOCONDUCTOR_CONFIG_FILE = "~/.bioconductor_config.yaml")
})
# this is important for the installation of package from R command
local({
    home_dir <- path.expand("~")
    conda_env_dir <- file.path(home_dir, "anaconda3/envs")
    set_envpath <- function(env_var, ...) {
        if (any(lengths(list(...)) > 1L)) {
            stop("Element of length one is expected", call. = FALSE)
        }
        add_path <- file.path(...)
        add_path <- path.expand(add_path)
        env_var_value <- Sys.getenv(env_var, unset = "")
        if (identical(env_var_value, "")) {
            path <- add_path
        } else {
            path <- paste0(add_path, switch(.Platform$OS.type,
                windows = ";",
                unix = ":"
            ), env_var_value)
        }
        expr <- quote(Sys.setenv())
        expr[[env_var]] <- path
        eval(expr)
    }

    # R-release environment path -----------------------------
    # conda create --name R-release -c conda-forge r-base git radian
    conda_renv <- file.path(conda_env_dir, "R-release")
    set_envpath("PKG_CONFIG_PATH", conda_renv, "lib", "pkgconfig")

    # pkg-specific PKG_CONFIG_PATH ---------------------------
    # these packages are not compitable with current conda R environment, install
    # this package in another conda environment and add the corresponding
    # PKG_CONFIG_PATH
    # gert environment variable
    # conda create -n libgit2 -c conda-forge libgit2
    libgit2 <- file.path(conda_env_dir, "libgit2")
    set_envpath("PKG_CONFIG_PATH", libgit2, "lib", "pkgconfig")
    set_envpath("LD_LIBRARY_PATH", libgit2, "lib")

    # textshaping environment variable
    # harfbuzz
    # conda create -n harfbuzz -c conda-forge harfbuzz
    harfbuzz <- file.path(conda_env_dir, "harfbuzz")
    set_envpath("PKG_CONFIG_PATH", harfbuzz, "lib", "pkgconfig")
    # fribidi
    # conda create -n fribidi -c conda-forge fribidi
    fribidi <- file.path(conda_env_dir, "fribidi")
    set_envpath("PKG_CONFIG_PATH", fribidi, "lib", "pkgconfig")
})

• .Renviron setting

LANGUAGE=en
R_HISTSIZE=500
# for I have no root permission, I use a user R library path
R_LIBS_USER=~/Rlibrary/%v
# add your conda R environment path to the `PATH` variable
PATH=/home/user/anaconda3/envs/R-release/bin:${PATH}

• last setting (optional), just for convenience of the usage of R in bash. .bashrc setting:

alias r="~/anaconda3/envs/R-release/bin/radian --r-binary ~/anaconda3/envs/R-release/bin/R"
alias R=~/anaconda3/envs/R-release/bin/R
alias Rscript=~/anaconda3/envs/R-release/bin/Rscript

[JiaxinLi-lipluszn]

Hi ! My environment is the same as raulguarini (OP): VsCode + R extension + SSH remote shell (Linux) + R only available through a Conda environment on the remote. I haven't tried to use radiant. After struggling a bit I just found how to use the Conda env properly, so I'm sharing it here. In my case, the Conda environment also contains R packages (tidyverse etc) and an ODBC driver that I need to use from R. In order to use syntax highlighting from VsCode, I installed r-languageserver and r-jsonlite in my Conda environment. Three options: the first one is a bit easier but not complete enough for me. I went for Option 3. I hope this can help someone else! If so, maybe document it somewhere? Pierre

Option 1

In the settings for the R extension, set RPath and the Rterm: (NB: I avoided "~" in the paths, not sure if expansion would work or not)

{
    "r.rpath.linux": "/home/myself/conda/envs/my_conda_env/bin/R",
    "r.rterm.linux": "/home/myself/conda/envs/my_conda_env/bin/R"
}

Result: R is available, tidyverse also OK but not the ODBC driver is not accessible from R.

Option 2

As suggested in #623, you can try to set "r.alwaysUseActiveTerminal": true in the VsCode settings. Then manually start a terminal, Conda-activate and start R. After that you should be able to submit R commands (e.g. with Ctrl+Enter). I find it less comfortable and didn't try it.

Option 3

Write a little shell script which first activates Conda and then starts R. I wrote it in /path/to/myproject/R-conda.sh and made it executable (chmod +x R-conda.sh)

#!/bin/bash
MY_CONDA_ENV=my_conda_env

# Optional and not related to Conda: cd to the script's dir
cd $(dirname $0)

# First we need to init conda bc ~/.bashrc is not executed when starting non-interactive shells. 
# Assuming PATH was already set by parent shell, and reusing script in profile.d/ if possible...
__conda_init_file=$(dirname $(which conda))/../etc/profile.d/conda.sh
if [ -f "$__conda_init_file" ]; then
    source "$__conda_init_file"
else
    eval "$(conda shell.bash hook 2> /dev/null)"
fi    

# Now we can activate
conda deactivate && conda activate $MY_CONDA_ENV

# And start R, passing through all command-line arguments
R "$@"

Then I pointed VsCode-R to this script in the settings:

{
    "r.rpath.linux": "/path/to/myproject/R-conda.sh",
    "r.rterm.linux": "/path/to/myproject/R-conda.sh"
}

Result: R is available, tidyverse and the ODBC driver are accessible from R.

Option 2 worked for me after conda installing r-languageserver and r-jsonlite - thanks so much for this post!! I like option 2 for constructing new code, but option 3 with command line args is probably better for debugging.

@ashrosewood Thanks a lot! Option 1 works for me! Though I didn't find rpath in the extension settings, only changing rterm and restarting VScode worked for me.

[Mohammadali-gharakhani]

Hi, I encountered a similar issue, but I resolved it by disconnecting from the remote extension host, which was on Linux, and then attempting to use the R extension