On Wed, Jun 08, 2016 at 01:13:10PM -0700, Father Chrysostomos via RT wrote: > On Wed Jun 08 11:06:18 2016, ikegami@adaelis.com wrote: > > On Tue, Jun 7, 2016 at 4:16 PM, Zefram <zefram@fysh.org> wrote: > > > > > In the fairly-common cases > > > where the comparator code is detectably innocuous, at least if it's just { > > > $a cmp $b } or { $a <=> $b }, we can still skip that copy. > > > > > > > Those aren't necessarily innocuous because of magic scalars and overloaded > > operators. > > Note that pp_sort already looks for magic and overloading at the outset and follows different code paths accordingly. Now (mostly) fixed by v5.25.3-188-g84721d6. -- My get-up-and-go just got up and went.Thread Previous