reallyTG/typethat

R Package for Dynamic Type Analysis

typethat

R Package for Dynamic Type Analysis

Jan 2019 // Work in Progress

Goals:

1. Analyze function usage in R packages;

2. Suggest type annotations for R functions;

3. Inform R compiler of function usage to ease early optimizations;

4. Inform design of a type system for R.

Usage:

For batch type analysis, start with the type_all_packages function:

type_all_packages(c("Rvmmin", "unpivotr")) # you may specify multiple packages

Output of the type analysis can be found in the type_res directory (which can be found in the working directory of the R instance that ran the previous command). You can then read individual result files with:

res_rvmmin <- readRDS("type_res/result_Rvmmin.RDS")
res_unpivotr <- readRDS("type_res/result_unpivotr.RDS")

Batch analysis of the output files is accomplished via the following:

# here, type_res is the directory containing the results
analysis_res <- analyze_all("type_res")

# to look at the "morphicity" of functions:
morphicity_counts <- analysis_res[[1]]

# to look at usage of more complicated functions:
polymorphic_functions <- analysis_res[[2]]

If profiles for single packages are of interest, consider the following functions:

# get output of one package
res_rvmmin <- readRDS("type_res/result_Rvmmin.RDS")

# get function profiles
rvmmin_profiles <- analyze_type_information(res_rvmmin)

# for a more readable version, try:
rvmmin_simplified <- simplify_analysis(rvmmin_profiles)

Note: R has several notions of "type" for a given value x, revealed through the typeof(x), class(x), mode(x), and storage.mode(x) functions. While our later analyses currently focus on the typeof notion of type, our initial type analysis does produce class, mode, and storage.mode information, which can be found in the files in the type_res directory.