Re: Proposed new test for Navigation Timing: Monolitic Clock

On Sat, 2012-12-08 at 11:47 -0500, Boris Zbarsky wrote:
> On 11/30/12 2:24 PM, Boris Zbarsky wrote:
> > On 11/30/12 1:32 PM, Philippe Le Hegaret wrote:
> >> Firefox fails on it: the difference between domainLookupStart and
> >> fetchStart is negative.
> >
> > I haven't been able to reproduce this so far, but I'll look into this.
> 
> I still haven't been able to reproduce.  Philippe, are you sure you're 
> seeing a negative domainLookupStart-fetchStart?

I just tried again and I don't seem to get it with Firefox nightly. I
get it sometimes with firefox 17 (Linux and Windows) if I force a
reload. So, it may well that it's not worth tracking it.

>   I do see a negative 
> connectStart-domainLookupEnd in some cases when reading from cache;

ditto.

>  I 
> filed https://bugzilla.mozilla.org/show_bug.cgi?id=819685 on that.

Thank you.

> One other note: Given DNS prefetchers, I'm not sure the restriction that 
> domainLookupStart must be >= various other things actually makes sense....
> 
> > I don't think you can assert that domLoading >= responseEnd as this test
> > does.  In fact, in the common case domLoading will be < responseEnd, I
> > would think.  Did you mean to test domInteractive >= responseEnd perhaps?
> 
> This is still a problem in the test.

Correct. I'm simply waiting for Jatinder to confirm that I need to
remove it. I'll ask him for domainLookupStart and DNS prefetchers as
well.

It may well be that, at the end, not much is left in the test to make it
interesting enough...

As a side note, I'm thinking to write a separate test for
secureConnection so we can track progress on that one separately.

Philippe

Received on Thursday, 13 December 2012 22:25:23 UTC