search

greta-dev / greta

simple and scalable statistical modelling in R
https://greta-stats.org
Other
529 stars 63 forks source link

Better installation checking #521

Open njtierney opened 2 years ago

njtierney commented 2 years ago

When using install_greta_deps(), it isn't picking up on errors when I pass them via callr.

This means I get something like this when I do a fresh install (note this is on branch #482 - which is working on getting TF 2.8.0 working with greta)

image

But then when I check with greta_sitrep() it does the following:

image

So perhaps I need some kind of validation step at each of the installation processes, that borrows from the parts of greta_sitrep()

njtierney commented 3 months ago

might be worthwhile revisiting

https://rstudio.github.io/reticulate/articles/versions.html

Config/reticulate:
  list(
    packages = list(
      list(package = "h5py", version = "2.8", pip = TRUE),
      list(package = "numpy", version = "1.16", pip = TRUE),
      list(package = "Pillow", pip = TRUE),
      list(package = "pip", version = "3.7", pip = TRUE),
      list(package = "python", version = "3.7", pip = TRUE),
      list(package = "pyyaml", pip = TRUE),
      list(package = "requests", pip = TRUE),
      list(package = "tensorflow-probability", version = "0.7", pip = TRUE),
      list(package = "tensorflow", version = "1.14", pip = TRUE)
    )
  )
njtierney commented 3 months ago

potentially revisit https://github.com/greta-dev/greta/issues/357 and https://rstudio.github.io/reticulate/articles/python_dependencies.html