reduce inference and optimization time (by turning it off)

[KristofferC]

Benchmarking the script in https://carlobaldassi.github.io/ArgParse.jl/stable/#Quick-overview-and-a-simple-example-1:

Master

❯ hyperfine --warmup=1 'julia --project script.jl foo'
Benchmark #1: julia --project script.jl foo
  Time (mean ± σ):      3.166 s ±  0.064 s    [User: 3.386 s, System: 0.287 s]
  Range (min … max):    3.075 s …  3.238 s    10 runs

PR:

❯ hyperfine --warmup=1 'julia --project script.jl foo'
Benchmark #1: julia --project script.jl foo
  Time (mean ± σ):      1.271 s ±  0.016 s    [User: 1.524 s, System: 0.264 s]
  Range (min … max):    1.253 s …  1.294 s    10 runs

[carlobaldassi]

👍 thanks

[KristofferC]

Would be nice with a new release whenever is suitable.

[oscardssmith]

Is there any other low hanging fruit for speeding this up (better precompile or something?)

[KristofferC]

You could experiment with SnoopCompile.jl.

[carlobaldassi]

It's released now.