Re: User Timing Mark Name for "Critical Content Loaded"?

On the Chrome side we are going to go ahead with implementing reporting for
the 'navigationLoaded' mark and do some evangelism around it.  It's
probably worth holding off on updating the spec until we see if there is
uptake from developers and how useful it ends up being.

On Thu, Jun 25, 2015 at 9:02 AM, Patrick Meenan <pmeenan@webpagetest.org>
wrote:

> Performance observers don't really give the browser (or performance tools)
> a better way to know about what the application developer cares about.
> They just give the developer a better way to track things (and maybe a
> better way to mark the "critical content loaded" point).  I think we still
> need an agreed-upon or convention for a mark name for Apps that do care to
> track it to expose it in a standard way.
>
> On Thu, Jun 25, 2015 at 7:57 AM, Philippe Le Hegaret <plh@w3.org> wrote:
>
>> On 06/24/2015 07:28 PM, Patrick Meenan wrote:
>>
>>> At least for what I was planning to do with it it wouldn't alter any
>>> behavior.  We (and I expect most browsers) track aggregate field metrics
>>> for a bunch of technical metrics to track our performance and guide our
>>> optimization work.  None of the standard technical measurements really
>>> mean anything for the user experience (onload, DOM Content Loaded,
>>> etc).  A lot of sites have their own custom metrics that they track that
>>> does better tie to the user experience and most that do have a core
>>> "this is the user experience time for this operation".  The time to
>>> first tweet and time to first pin were concrete examples that I know of
>>> but just about every major web property has their own.
>>>
>>> What I'd like to do is to be able to collect that in a standard way so
>>> that when we make optimization trade-offs we take the applications
>>> actual experience metrics into account.  That does mean that it will
>>> impact decisions that we make about how the browser works but not in the
>>> context of that specific page or page load.
>>>
>>
>> With performance observers, this would give you an easier to track those
>> marks, correct?
>>
>> Philippe
>>
>>
>>
>