Ondrap/master

[tmcgilchrist]

Rebased version of https://github.com/liyang/thyme/pull/45.

Needs further evaluation of:

1. the scanner dependency, should we pickup an additional dependency and is the package well enough maintained to choose.
2. Run more benchmarks on recent GHC versions and compare against attoparsec

Benchmarks on macOS 12.6.3, GHC 8.10.7, 2.3 GHz Quad-Core Intel Core i7:

Running 1 benchmarks...
Benchmark bench: RUNNING...
benchmarking time encode/time/encode
time                 741.9 ns   (731.2 ns .. 751.0 ns)
                     0.999 R²   (0.999 R² .. 1.000 R²)
mean                 730.0 ns   (726.4 ns .. 736.7 ns)
std dev              15.80 ns   (10.20 ns .. 22.96 ns)
variance introduced by outliers: 27% (moderately inflated)

benchmarking time encode/thyme/encode
time                 353.7 ns   (350.0 ns .. 358.8 ns)
                     0.999 R²   (0.998 R² .. 0.999 R²)
mean                 358.7 ns   (354.5 ns .. 364.7 ns)
std dev              17.09 ns   (13.41 ns .. 21.56 ns)
variance introduced by outliers: 66% (severely inflated)

benchmarking time decode/time/decode
time                 1.465 μs   (1.462 μs .. 1.470 μs)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 1.465 μs   (1.463 μs .. 1.468 μs)
std dev              7.874 ns   (5.819 ns .. 10.35 ns)

benchmarking time decode/thyme/decode
time                 617.2 ns   (612.0 ns .. 624.1 ns)
                     1.000 R²   (0.999 R² .. 1.000 R²)
mean                 614.4 ns   (612.7 ns .. 618.0 ns)
std dev              8.069 ns   (3.528 ns .. 14.65 ns)
variance introduced by outliers: 12% (moderately inflated)

Benchmark bench: FINISH

[tmcgilchrist]

Closing in favour of haskell-github-trust#75