W3C home > Mailing lists > Public > public-web-perf@w3.org > November 2010

Re: [NavigationTiming] a couple questions

From: Jonas Sicking <jonas@sicking.cc>
Date: Tue, 9 Nov 2010 23:26:52 -0800
Message-ID: <AANLkTi=A80EYLYeEkszdHOByp+ALtK+U=0+OLinHttfj@mail.gmail.com>
To: Zhiheng Wang <zhihengw@google.com>
Cc: public-web-perf <public-web-perf@w3.org>
On Tue, Nov 9, 2010 at 10:35 PM, Zhiheng Wang <zhihengw@google.com> wrote:
>
>
> On Tue, Nov 9, 2010 at 10:11 PM, Jonas Sicking <jonas@sicking.cc> wrote:
>>
>> On Tue, Nov 9, 2010 at 9:47 PM, Zhiheng Wang <zhihengw@google.com> wrote:
>> >     Jonas brought up a couple questions about the current draft and we
>> > decide to
>> > have them discussed here.
>> > - "One thing that is a bit unclear to me is the navigationStart
>> > attribute. In one of the two figures it is indicated to be available
>> > on same-origin only navigation. But in the other figure, and in the
>> > description of the attribute itself, it seems like it's available on
>> > cross-origin navigation too."
>> >    The first figure doesn't include redirect, in which case
>> > navigationStart
>> > is available all the time regardless the origins. Meanwhile,
>> > unloadEventStart
>> > and unloadEventEnd only reveal under the same origin. (They should have
>> > been underlined in the first figure. I will fix it in the next update.)
>>
>> So what is the difference between navigationStart and fetchStart for
>> requests without redirects? Is it just the delay in the browsers
>> network stack due to for example other in-progress requests?
>
>    Something like that. You are right that the difference is minor at this
> time.
> It does exist though and no telling how that changes over time. So the idea
> is to getting to the user action as close as possible.

Cool.

>> >    The recent discussion on SOP has been very educating. :-)
>>
>> I wish it was simpler :(
>>
>> > - "Also, an issue that isn't related to privacy. This introduces a new
>> > global attribute called 'performance', right? This seems somewhat
>> > scary as it is quite a generic name, and will break any web pages that
>> > uses a global variable with that name."
>> >   hmm, any other name to suggest?
>>
>> I'm trying to stay away from naming discussions (for the most part).
>>
>> One solution would be to stick it on the document object. Another to
>> come up with a more exotic name though I have to admit any good
>> suggestions escape me at the moment.
>
>   Sounds like another topic for tomorrow's conf call. Care to join us?

Sorry, don't think I'll have time to make that :(

/ Jonas
Received on Wednesday, 10 November 2010 07:27:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 21 December 2010 18:13:55 GMT