808102 - kraken-parse-financial regression Mac OS X 32b (nov 2, delta 5ms)

Status: RESOLVED INVALID

(Core :: JavaScript Engine, defect)

Description

[Nicolas B. Pierron [:nbp] {back on 17/11}]

AWFY reported on second regression on kraken-parse-financial.  Here is the list of changeset which have been touching JS files between the 2 changesets:

* e89f3933239117e7cadc2dc7618aaf4bb78f1b50 - Bug 807237. Add 'data' parameter to JS_StealArrayBufferContents. r=sfink <Robert O'Callahan>
* 4d0cfc4fd49f0c7356cd34af8188002089a44957 - Bug 807212. Call TypedArray methods instead of accessing slots directly. r=sfink <Robert O'Callahan>
* 95187d8e170c6df98ca83267a65667a9bff92118 - Bug 807223 - Implement JS_ArrayBufferViewBuffer. r=sfink This is really sfink's patch, but we wrote basically the same code and the review looks better this way. <Robert O'Callahan>

Comment 1

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I guess it can only be "Call TypedArray methods instead of accessing slots directly" because the other patches would not even be touched by kraken. But that patch shouldn't really change anything .. could it be a failure to inline TypedArray methods?

Comment 2

[Jan de Mooij [:jandem]]

See bug 808098 comment 2, and Kraken doesn't use typed arrays at all. parse-financial just benchmarks the JSON parser.

Comment 3

[Robert O'Callahan (:roc) (email my personal email if necessary)]

OK, any regression can't be those patches then.

Comment 4

[Steve Fink [:sfink] [:s:]]

Sounds like this might just be noise in the measurements to begin with. Should this be marked INVALID? (Is there anything we can do about controlling the measurement variance? Might be worth another bug.)