Re: [NavigationTiming] Move window.performance to navigator.performance

On Fri, 10 Dec 2010 00:07:19 +0100, Anderson Quach <>  

> Hi Folks,
> In the meeting on 12/1/2010  
> [minutes<>], as a  
> working group we decided to go with window.performance as the official  
> namespace for the Web Performance Working Group. We reviewed the risks  
> of this name and its alignment to the working group's charter.
> The decision was based on a number of different factors:
> i.                     The namespace performance is much more suitable  
> given the direction we want to take with this working group in terms of  
> helping site operators and site developers access and understand their  
> performance in user agents. Particularly as we move forward with adding  
> functionality to this interface. The alternate names we considered were  
> pagePerformance and webPerformance.
> ii.                   The risks as outlined in Zhiheng's response and in  
> our previous study on high traffic sites were found to be low for the  
> name "performance".
> iii.                  When analyzing the occurrences of pagePerformance,  
> the analysis came back inconclusive, no concrete instances of "var  
> pagePerformance" or "var webPerformance" were found.

Did you look for occurances of name="performance" and id="performance" in  
HTML? Those also end up as properties on the global object. (I included a  
link to google code search for name="performance" in )

Note that 'var foo' would win over the native, so it wouldn't  
break the scripts of such pages (but it would not be possible to use the  
native since it's overwritten). However, the native  
wins over name="foo" and id="foo", so it's possible to break scripts that  
assume that returns an element or HTMLCollection. Maybe there  
aren't high-traffic sites that expect that, but there's a very long tail  
of Web pages out there that rely on the silliest things...

> Best Regards,
> Anderson Quach
> IE Program Manager
> From:  
> [] On Behalf Of Zhiheng Wang
> Sent: Tuesday, November 23, 2010 2:50 PM
> To: Jonas Sicking
> Cc: Simon Pieters; Bryan McQuade;
> Subject: Re: [NavigationTiming] Move window.performance to  
> navigator.performance
>   So "pagePerformance" sounds good to everyone?
>   Nic and Anderson found that 0.00993% of high traffic websites has  
> performance object in
> their scripts. It might not cover all the cases, but I sort of subscribe  
> to the idea that these
> developers should know about this draft, especially if they care about  
> "performance". };->
> That said, using a rarer name helps.
> cheers,
> Zhiheng
> On Tue, Nov 23, 2010 at 12:05 PM, Jonas Sicking  
> <<>> wrote:
> On Tue, Nov 23, 2010 at 11:58 AM, Simon Pieters  
> <<>> wrote:
>> On Tue, 23 Nov 2010 19:34:54 +0100, Bryan McQuade  
>> <<>>
>> wrote:
>>> Agree that the property name performance is likely to collide with
>>> existing
>>> scripts. Maybe we need to choose a different less common name?
>> That works for me. Any suggestions? "pagePerformance" doesn't come up  
>> much
>> in google code search.
> Sold!
> / Jonas

Simon Pieters
Opera Software

Received on Friday, 10 December 2010 00:06:23 UTC