Observed that parsing is slower that it needs to be, due to the map lookups:
In this PR I replace double lookup (containsKey + get) with simple array access.
I did not benchmark properly but simply ran testDecode 100x, several times
Before: ~8s
After: ~6s
Observed that parsing is slower that it needs to be, due to the map lookups:
In this PR I replace double lookup (containsKey + get) with simple array access. I did not benchmark properly but simply ran
testDecode
100x, several times Before: ~8s After: ~6s