R aws access

[wildintellect]

Troubleshooting AWS S3 access from R we discovered that the current library we install aws.s3 (aka r-aws.s3) is unmaintained, and that other alternatives are equally challenging s3 (not in conda-forge).

Reading up, AWS recommends a package called paws (aka r-paws) which is an R SDK similar to the Python AWS SDK.
So you can do things very much like boto3, however it's not very R styled.

Ask

• Update the R image to include paws, possibly remove aws.s3 <- some user code currently uses this though s3_get specifically (Platform Team)
• Update Docs (Data Team) with examples

Example:

Get a list of files from s3

#bash
conda install -c conda-forge r-paws

#R
library(paws)
s3 <- paws::s3()
results <- s3$list_objects_v2(Bucket = "maap-ops-workspace", Prefix="shared/abarenblitt/GEDI_global_PA_v2/")
file_list <- sapply(results$Contents, function(x) {x$Key})

thanks to @abarenblitt for reporting issue
cc: @smk0033

[grallewellyn]

How do we want to determine if we should remove r-aws.s3 from the R image?

[wildintellect]

For now it's not a conflict to also install paws so I would start with that. We have an open ticket to write best practices with R, we'll make sure not to use the likely to be removed library.