Lame. We have a bug in reduce/reduceRight then because we use this pattern which is observable.
Maybe %HasFastPackedElements is fine after all since that will be false if we have a getter. Need to test a bit more. On Fri, Aug 9, 2013 at 11:40 AM, <ad...@chromium.org> wrote: > On 2013/08/09 14:24:06, arv wrote: > >> One more idea... taken from reduce >> > > Instead of doing the in check we [[Get]] the value and only if the value >> is >> undefined do we need to do an [[HasProperty]] check. >> > > ForEach: 184870 >> ForEach2: 367066 >> ForEach3: 452772 >> ForEachWithHoles: 123503 >> ForEach2WithHoles: 114547 >> ForEach3WithHoles: 110123 >> > > This is 3x faster! >> > > The downside to both of these ideas is that with proxies it is detectable >> that >> the order no longer matches the spec. >> > > It's not just proxies, it's also observable with getters. If you look at > git > blame, that's what the code used to do, and I think mstarzinger fixed it to > match the spec. > > > Maybe these hacks are not worth it and we should just make [[HasProperty]] >> fast... not that at will ever be as fast as ForEach3 since we always need >> to >> > do > >> the [[Get]] no matter what. >> > > > > https://codereview.chromium.**org/22545007/<https://codereview.chromium.org/22545007/> > -- erik -- -- v8-dev mailing list v8-dev@googlegroups.com http://groups.google.com/group/v8-dev --- You received this message because you are subscribed to the Google Groups "v8-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to v8-dev+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
