Improve speed of group_by()

[etiennebacher]

Related to #119.

Benchmark setup:

d <- data.frame(
  g1 = sample(LETTERS, 8000, TRUE),
  g2 = sample(LETTERS, 8000, TRUE),
  g3 = sample(LETTERS, 8000, TRUE),
  x1 = runif(8000),
  x2 = runif(8000),
  x3 = runif(8000)
)

# return a list of results so that both functions return the same output (without
# all the class problem, tibble vs data.frame)
poor = function() {
  foo <- d %>% 
    group_by(g1, g2, g3) |> 
    summarise(x1 = mean(x1), x2 = max(x2), x3 = min(x3))

  list(foo$x1, foo$x2, foo$x3)
}

dpl = function() {
  foo <- d %>% 
    dplyr::group_by(g1, g2, g3) |> 
    dplyr::summarise(x1 = mean(x1), x2 = max(x2), x3 = min(x3))

  list(foo$x1, foo$x2, foo$x3)
}

bench::mark(
  poor(),
  dpl()
)

Before:

#> # A tibble: 2 × 13
#>   expression      min   median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result     memory     time           gc      
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list>     <list>     <list>         <list>  
#> 1 poor()        11.6s    11.6s    0.0863    1.83GB    11.0      1   127      11.6s <list [3]> <Rprofmem> <bench_tm [1]> <tibble>
#> 2 dpl()       157.2ms  178.3ms    5.77      7.34MB     5.77     3     3    520.2ms <list [3]> <Rprofmem> <bench_tm [3]> <tibble>

After:

#> # A tibble: 2 × 13
#>   expression      min   median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result     memory     time           gc      
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list>     <list>     <list>         <list>  
#> 1 poor()        7.77s    7.77s     0.129    1.83GB     9.78     1    76      7.77s <list [3]> <Rprofmem> <bench_tm [1]> <tibble>
#> 2 dpl()      112.42ms 152.17ms     6.66     3.39MB     3.33     4     2   601.02ms <list [3]> <Rprofmem> <bench_tm [4]> <tibble>

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.60%. Comparing base (3cc0a99) to head (e6e28e8).
⚠️ Report is 7 commits behind head on master.
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #121      +/-   ##
==========================================
+ Coverage   93.59%   93.60%   +0.01%     
==========================================
  Files          59       59              
  Lines        1483     1486       +3     
==========================================
+ Hits         1388     1391       +3     
  Misses         95       95              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[nathaneastwood]

I've been away. I had every intention of reviewing this this evening but I spent it catching up on other things. I'll try and look tomorrow after work.