Support for Python package environment on shinyapps

Posit Professional & Hosted Products shinyapps.io
1

I am attempting to optimize a R shiny application on shinyapps. The issue arises due to the use of reticulate and 5 python packages that the application needs to function. I have followed the steps outlined in: https://github.com/ranikay/shiny-reticulate-app. However, although I have the application working with this method, it is quite inefficient as the user needs to wait around 90 seconds for the packages to download with:

virtualenv_dir = Sys.getenv('VIRTUALENV_NAME')
python_path = Sys.getenv('PYTHON_PATH')
  
# Create virtual env and install dependencies
reticulate::virtualenv_create(envname = virtualenv_dir, python = python_path)
reticulate::virtualenv_install(virtualenv_dir, packages = PYTHON_DEPENDENCIES,  ignore_installed=TRUE)
reticulate::use_virtualenv(virtualenv_dir, required = T)

Is there any way to install python packages at the deployment phase of releasing a R shiny application. Ideally a way that includes the use of a requirements.txt file that allows creation of the environment at deployment, similarly to Shiny for Python https://shiny.posit.co/py/docs/deploy.html#create-a-requirements.txt. Any solution that doesn't involve the user waiting for python package installation upon application 'wake-up' would be fantastic.

Any help with this would be greatly appreciated!

2

I realize that this is an old request, but I am running into the same issue. And according quick google search, I am not alone.

Having to install a venv along with the python packages needed at runtime is an enormous drag and makes pursuing entire classes of online apps unfeasible. I have tried many workarounds for my use case, none have worked.

Is there any hope that you will make venvs persistent between sessions? If not, I and a lot of other people will have to abandon RStudio for these apps.