Implementing the fastfunc itself did shave some time off the t/fn-read.t test, so there's that. A further idea of making the fast string object "stateful" (keeping a position which could be unsafely updated by the fastfunc) did not make a significant difference.
Implementing the fastfunc itself did shave some time off the t/fn-read.t test, so there's that. A further idea of making the fast string object "stateful" (keeping a position which could be unsafely updated by the fastfunc) did not make a significant difference.