Closed thufschmitt closed 6 months ago
I'm not entirely clear why, but this shoves off 20s (60%) of the evaluation time.
$ hyperfine --warmup=1 -L REF HEAD,HEAD^ 'nix develop --option eval-cache false git+file://$PWD\?rev=$(git rev-parse {REF}) --command true' Benchmark 1: nix develop --option eval-cache false git+file://$PWD\?rev=$(git rev-parse HEAD) --command true Time (mean ± σ): 11.693 s ± 1.475 s [User: 4.940 s, System: 1.424 s] Range (min … max): 10.723 s … 14.223 s 10 runs Benchmark 2: nix develop --option eval-cache false git+file://$PWD\?rev=$(git rev-parse HEAD^) --command true Time (mean ± σ): 32.296 s ± 1.090 s [User: 15.727 s, System: 3.563 s] Range (min … max): 31.281 s … 34.447 s 10 runs Summary nix develop --option eval-cache false git+file://$PWD\?rev=$(git rev-parse HEAD) --command true ran 2.76 ± 0.36 times faster than nix develop --option eval-cache false git+file://$PWD\?rev=$(git rev-parse HEAD^) --command true
I'm not entirely clear why, but this shoves off 20s (60%) of the evaluation time.