On Mon, May 06, 2002 at 09:38:00PM -0700, Brent Dax wrote: > (FYI, I'm the main author of Parrot's rx package.) > > Mark Kvale: > # computes dynamically several functions, such as length of the > # regex match string, that are precomputed in the reg code... > # > # Note that the overhead/match ratio lo/ro is large in the reg, > # spen and perl routines. In the reg code, I think this is due > # to the allocation of two PerlArrays for subexpression > # indices. In the spen and perl code, it is due to data > # structure initialization and optimization checks. > > It may be disabled in the current code, but rx allocates a pair of > PerlArrays too.
It's still there, in rx_clearinfo.