From Microsoft's paper on their new derivative-based regex engine (RegexOptions.NonBacktracking):
Incidentally, the result for Rust further highlights the importance of avoiding outliers, as it would be a clear winner if not for the pattern [a-q][ˆu-z]{13}x. The crux of the pattern is that [a-q] denotes a subset of [ˆu-z] much like in the classical example a.{13}. This pattern is challenging for DFA based engines, as the loop can be entered multiple times, leading to a 2^13 factor in number of states. Optimizations in NonBacktracking have made this pattern no longer visible as an outlier.
It might be worthwhile including this case in the benchmarks.
From Microsoft's paper on their new derivative-based regex engine (
RegexOptions.NonBacktracking):It might be worthwhile including this case in the benchmarks.