Anyone have any idea why, in v8, a function that does `foo[bar]` on an identical (identity) object would go megamorphic?
@mraleph the inputs are stable types and so are the outputs...
-
-
@wycats why do you think it goes megamorphic? -
@mraleph The assumption was based on getting the "optimized too many times" error in the profiler. -
@wycats ... it's just means V8's optimizations do not stabilize - often this means a bug in V8 itself -
@mraleph Interesting. I'll try again on beta and try to use IRHydra more to investigate more deeply. -
@wycats ping me if you have any questions either here or me@mrale.ph I'll be away this evening but I'll reply as soon as I am back. -
@mraleph thanks for pointing me to IRHydra. It's very very helpful. -
@wycats yay, I am glad it helps! [DevTools folks recently started working on surfacing similar deopt data right in their UI.] - 2 more replies
New conversation -
-
-
@wycats ... deopt trying convert string to an integer, on newer channels - this is fixed.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@wycats if you look at deops in IRHydra you would see a picture like this (for stable Chrome)pic.twitter.com/W0CsbzyBGN
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@wycats if foo is of the same hidden class and bar is the same string it should not go megamorphic unless there is a bug in V8Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.