benchmarking ghc (`elem` "{([") vs. (\c -> c `elem` "{([") vs. ...

string_vs_list.R

library(pacman)
p_load(tidyverse, tictoc, glue, e1071, directlabels, cache)

split <- function(x) unlist(strsplit(x, ""))

sample(c(letters, split("{}[]()")), 10^5, replace = TRUE) %>% as.list() %>% do.call(str_c, .) %>% writeLines("input0.txt")


# map_chr(commands, function(command) readLines(paste0(command, ".hs"))) %>% setNames(commands) %>% deparse %>% writeLines
mains <- c(
  "./listEtaEx" = "main = print . length . filter (\\c -> c `elem` ['{', '(', '[']) =<< readFile \"input.txt\"",
  "./list" = "main = print . length . filter (`elem` ['{', '(', '[']) =<< readFile \"input.txt\"",
  "./string" = "main = print . length . filter (`elem` \"{([\") =<< readFile \"input.txt\"",
  "./stringEtaEx" = "main = print . length . filter (\\c -> c `elem` \"{([\") =<< readFile \"input.txt\""
)

iwalk(mains, ~ {
  writeLines(.x, glue("{.y}.hs"))
  system(glue("ghc -O2 {.y}.hs"))
})

commands <- Sys.glob("*.hs") %>% str_remove(".hs$") %>% paste0("./", .)

pmat <- permutations(length(commands))

cache(tictocs =
  map_dfr(log(seq(exp(5), exp(7), length.out = 7)), function(b) {
    system(glue("cat {paste(rep('input0.txt', round(10^ (b-5))), collapse = ' ')} > input.txt"))
      map_dfr(1, function(replicate) {
        map_dfr(sample(nrow(pmat), 10), function(permu) {
          map_dfr(pmat[permu, ], function(i) {
            tic()
            system(commands[i])
            r <- toc()
            time <- with(r, toc - tic)
            tibble(time = time, com = commands[i], b = b, n = 10^b, replicate = replicate)
          })
        })
      })
    })
)

tictocs <- within(tictocs, {
  str <- str_detect(tictocs$com, "string")
  etaEx <- str_detect(tictocs$com, "EtaEx")
  opts <- ifelse(str, "string", "list")
  label <- ifelse(etaEx, "\\ c -> c `elem` _", "(`elem` _)")
})

plot(
  ggplot(tictocs, aes(b, time, label = label, group = com)) +
    geom_point(aes(col = opts)) +
    theme_minimal() +
    geom_dl(method = "last.qp", cex = 8) +
    geom_smooth(aes(col = opts)) +
    scale_color_manual(name = "_ =", values = c("red", "blue")) +
    xlim(5, 8)
)