You can clone with
Encoding performance: compare like with like.
Make benchmarks more comprehensive.
Make the benchmarks fairer.
Build Haskell benchmarks with cabal
encode and decode JSValues instead of JSObjects
This is consistent with the aeson encode en decode.
This does not change the benchmark results.
Added CompareWithJSON benchmark from issue: #56