Re: navigationStart in NavigationTiming

I mean to say the behavior is unspecified for "about:blank", not undefined.
Arvind


On Wed, May 1, 2013 at 6:10 PM, Arvind Jain <arvind@google.com> wrote:

> I see. Given that the specification is about "navigations" e.g. see the
> abstract or the processing model, do we then need to define anything for
> pages that are only "created"? It appears we will need to make a bunch of
> updates to the spec to cover the case of a document that is not navigated
> to and just "created". For example, we need to update the processing model,
> and we need to update the navigation type (
> http://www.w3.org/TR/navigation-timing/#performancenavigation).
>
> Is it better to say in a non-normative section that performance.timing and
> performance.navigation are undefined for "about:blank"?
>
> Arvind
>
>
> On Wed, May 1, 2013 at 5:49 PM, Boris Zbarsky <bzbarsky@mit.edu> wrote:
>
>> On 5/1/13 5:48 PM, Arvind Jain wrote:
>>
>>> If there is no previous document and the current
>>> document is "about:blank", this attribute must return the time when the
>>> document was created.
>>>
>>
>> I don't think we need the "the current document is 'about:blank'" bit,
>> even.  Any case when there is no previous document means the document is
>> "about:blank".
>>
>>
>>  I actually feel it's better to just not mention the case of "there is no
>>> previous document" and current document = about:blank.
>>>
>>
>> If we don't mention it, what defines the behavior?
>>
>>
>>  Please also look at the processing model for navigation here:
>>> http://www.w3.org/TR/html5/**browsers.html#navigate<http://www.w3.org/TR/html5/browsers.html#navigate>- it simply assumes
>>> there is a previous document (Step 9)
>>>
>>
>> When _navigating_ there is a previous document in all cases.  But the
>> initial about:blank is not navigated to; it's just created.  But it has a
>> performance.timing, and we need to define what the values in that are.  We
>> can define them to be creation time, or all 0, or whatever, but we need to
>> define them.
>>
>>
>>  What would be the ideal definition of these 3 attributes according to
>>> you?
>>>
>>
>> I don't actually have a strong opinion about how we should define the
>> behavior of performance.timing for the initial about:blank.  It just needs
>> to be defined.
>>
>> -Boris
>>
>>
>