Here I added a bit of simple logic to avoid running renv/activate.R when we are in a Docker image. To do this I added an ENV variable named OPENSCPCA_DOCKER to all docker images, and then only source the activate script if the value is not set to TRUE.
The original issue isn't so much that related to caching in the docker images as that the files are being installed in the system library, so we don't want renv to be activated and looking for its own versions of the packages. If we were to do that, we would almost certainly run into path problems with files stored or not in the host filesystem.
I tested this in the hello-R image/script, but applied the same changes everywhere, including in the create-analysis-module.py script to cover future modules. I also included the ENV variable in the template, and in the hello-python module Dockerfile, even though it may not be needed there. Seemed a good place for just a bit of extra consistency.
Closes #542
Here I added a bit of simple logic to avoid running
renv/activate.Rwhen we are in a Docker image. To do this I added an ENV variable namedOPENSCPCA_DOCKERto all docker images, and then only source the activate script if the value is not set toTRUE.The original issue isn't so much that related to caching in the docker images as that the files are being installed in the system library, so we don't want renv to be activated and looking for its own versions of the packages. If we were to do that, we would almost certainly run into path problems with files stored or not in the host filesystem.
I tested this in the hello-R image/script, but applied the same changes everywhere, including in the
create-analysis-module.pyscript to cover future modules. I also included the ENV variable in the template, and in the hello-python module Dockerfile, even though it may not be needed there. Seemed a good place for just a bit of extra consistency.