There are a few alternatives to RDS for data frames like parquet or fst. However, for saving generic R objects, I was only able to find qs (being replaced by qs2). Compared to RDS files, qs ones are substantially faster to read and write. Although I have not noticed any problems, I am hesitant to use qs in my workflows. Is there a reason this alternative is not more widely adopted?
If there are ways to optimize reading/writing R objects, why are they not incorporated into the native RDS format? There are multiple versions of RDS, so it has been tweaked in the past.