fmisc

Various utilities for software development in R: composable function decorators, memory-aware chunked processing, smart parallel apply, rich error messages, and package-development helpers.

Features

• Function decorators: stackable function operators — retry with exponential backoff (with_retry()), timing (with_timing()), logging (with_logging()), memoisation (with_cache()), and rate limiting (with_rate_limit()). Compose with decorate(), inspect with is_decorated(), unwrap with undecorate().
• Chunking & RAM management: process data larger than memory via process_with_chunks(), with automatic chunk sizing, cross-platform RAM monitoring (Rcpp-accelerated), and disk spillover.
• Smart parallel computing: smart_parallel_apply() picks an OS-appropriate backend automatically and falls back to sequential processing when no backend is available.
• Better errors: stop2() produces rich, {interpolated} error messages via cli/rlang with a base-R fallback.
• Development tools: a best-practices function template (use_function_template()) and a package Makefile (use_make2()).

Installation

You can install the development version of fmisc from GitHub with:

# install.packages("pak")
pak::pak("fabiandistler/fmisc")

Function Decorators

library(fmisc)

flaky <- function(x) {
  if (runif(1) < 0.7) stop2("transient failure")
  x * 2
}

robust <- flaky |>
  with_retry(max_tries = 5, backoff = 0.1) |>
  with_logging()

robust(21)

Chunking & RAM Management

big_df <- data.frame(x = rnorm(1e7), y = runif(1e7))

result <- process_with_chunks(
  big_df,
  process_fn = function(chunk) colMeans(chunk),
  combine_fn = function(parts) rowMeans(do.call(rbind, parts)),
  max_ram_mb = 512,
  verbose = FALSE
)

Smart Parallel Computing

results <- smart_parallel_apply(1:1000, function(i) sqrt(i))

Development Tools

Function Template

Create new R functions from a comprehensive best-practices template:

library(fmisc)

# Create a new function file with tidyverse-style template
use_function_template("my_function")

The template includes:

• Roxygen2 documentation patterns
• Argument ordering checklist (data → descriptors → … → details)
• Multiple validation options (stopifnot, cli, rlang, checkmate)
• Dots handling patterns
• Error handling with custom error constructors
• Design principles checklist

Makefile for R Packages

Add a comprehensive Makefile for package development:

# Add Makefile with common targets
use_make2()

# Then use make commands in your terminal
# make all      - document, build, and check
# make test     - run tests
# make help     - see all available targets