Discover this year's submissions to the Dataiku Frontrunner Awards and give kudos to your favorite use cases and success stories!READ MORE

Adding bitbucket library in code environment UI

ak12
Neuron
Neuron
Adding bitbucket library in code environment UI

I am looking to add an R library in a private repo in bitbucket to a code env, I have tested adding it as a project level library as per this document (https://knowledge.dataiku.com/latest/courses/governance/collaboration/git-projects.html) which is working fine.

However ideally I would have the package in the code env so it is more widely available. I have successfully installed packages in the code env from github using this syntax: git+https://github.com/<package>.git but I cannot seem to get it to work with a bitbucket link. The error I am getting just says that it is not available. For reference I am using Dataiku version 9.0.5

Is there anything I am overlooking? Or is this only available through github?

0 Kudos
2 Replies
AlexT
Dataiker
Dataiker

Hi @ak12 ,

Is your bitbucket repo public or private? If it's private repo you may need to create an app password:

https://support.atlassian.com/bitbucket-cloud/docs/app-passwords/

And include this in your URL: 

git+https://USER_NAME:APP_PASSWORD@GIT_URL/PATH_TO_YOUR_REPO.git

 

Thanks,

0 Kudos
sergeyd
Dataiker
Dataiker

Hi, 

If we are talking about R code envs, the quick answer is no, it's not possible to install R packages from the git repositories from DSS UI.

In R, to install the package from Git, you need to do this manually from the CLI using either "devtools" or "remotes" R packages. This will require something like this: 

install.packages('devtools')
require(devtools)
devtools::install_github("dmlc/xgboost")
# or:
library(devtools)
install_github("dmlc/xgboost")
which is not possible from DSS UI as we search for packages only within the R CRAN mirror set in the code env config. 
 

The syntax for the package:

git+https://github.com/<package>.git
works for python code env as you can use this syntax in pip install commands but this doesn't work for R.