My C++ time stamp resolution was misleading. All is well after all. I should have known it was me and sending the mail late was a bad omen.
Sorry for wasting your time. On Tuesday, September 11, 2012 2:45:47 AM UTC-4, Matthias Ernst wrote: > > On Tue, Sep 11, 2012 at 7:19 AM, Stuart Allen > <[email protected]<javascript:>> > wrote: > > Hi, > > > > I stumbled across a very strange problem. I'm using 3.12.19.1 on Linux > > 64-bit. > > > > I have an object implemented in C++, I noticed an unexpected slow-down > and > > eventually traced it to a particular call. > > > > If the object method is called like this: > > > > this.MyObject.method() > > > > > > it takes 20ms (milliseconds) to execute, when it should be 0ms. I made > > timings inside C++ and they show a total 0ms execution time so the 20ms > is > > introduced inside the JS<->C++ boundary. > > > > On a hunch I tried this: > > > > var cn= this.MyObject.method; > > cn(); > > For the record, this call has different semantics: it won't pass > MyObject, but rather the global object, as "this". Not that that would > immediately explain the timing difference you see. > > I think you may need to post some more context code. > Especially regarding the construction of "MyObject" including "method". > > Is the timing you see measured as an average of many loop iterations > or just one call? > > Matthias > > > > > > > and guess what, it takes 0ms!! > > > > methods of the same Object instance are called before and after yet they > > execute as expected with no measurable overhead (in millsecond land). > > > > Is there any foundation for such a problem? Could it be a > scoping/resolving > > issue causing problems resolving the method? > > > > Are there any diagnostics that could help here? > > > > Thanks, > > > > Stuart. > > > > -- > > v8-users mailing list > > [email protected] <javascript:> > > http://groups.google.com/group/v8-users > -- v8-users mailing list [email protected] http://groups.google.com/group/v8-users
