The following revision seems to have regressed the lastIndexOf subtest by 2.5-4x: changeset: 26344:1c6be1c210b9 user: Andreas Gal <email@example.com> date: Fri Mar 20 18:52:11 2009 -0700 files: js/src/jsapi.h js/src/jsfun.h js/src/jstracer.cpp js/src/jstracer.h description: Support calling arbitrary JSFastNatives from trace (463238, r=brendan).
maybe bug 483723 would help this?
4x is a lot. Any idea where exactly the slowdown is coming from? We used to have a slightly more efficient builtin there, but that shouldn't make a 4x difference.
I don't know the cause; right now I'm just trying to get the full list of Dromaeo regressions so we know what's going on. I found that this regression also affects the 'substr' and 'indexOf' subtests.
David, do you know whether this is still an issue?
Closing out old regressions that have since been eclipsed by gains and overtaken by rearchitecture.