RE: [Resource Timing] Why does connectEnd exclude the SSL Handshake?

I've updated the Resource Timing connectEnd definition to be more consistent with the Navigation Timing connectEnd definition, https://dvcs.w3.org/hg/webperf/raw-file/tip/specs/ResourceTiming/Overview.html#dom-performanceresourcetiming-connectend.

Thanks,
Jatinder

From: Arvind Jain [mailto:arvind@google.com]
Sent: Wednesday, April 10, 2013 9:15 AM
To: Nic Jansma
Cc: Austin,Daniel; James Simonsen; Andy Davies; public-web-perf@w3.org
Subject: Re: [Resource Timing] Why does connectEnd exclude the SSL Handshake?

Yes let's fix it. I suspect it's just an oversight - we changed the text in Navigation Timing as a result of this thread:
 http://lists.w3.org/Archives/Public/public-web-perf/2010Nov/0046.html
and we probably forgot to make the change in Resource Timing specification.


On Wed, Apr 10, 2013 at 8:05 AM, Nic Jansma <nic@nicj.net<mailto:nic@nicj.net>> wrote:
NavigationTiming and ResourceTiming differ in how connectEnd is defined:

NavigationTiming (http://www.w3c-test.org/webperf/specs/NavigationTiming/):
connectEnd<https://dvcs.w3.org/hg/webperf/raw-file/tip/specs/NavigationTiming/Overview.html#dom-performancetiming-connectend> must include the time interval to establish the transport connection as well as other time interval such as SSL handshake and SOCKS authentication.
ResourceTiming (http://www.w3c-test.org/webperf/specs/ResourceTiming/):

connectEnd<http://www.w3.org/TR/resource-timing/#dom-performanceresourcetiming-connectend> must include the time interval to establish the transport connection. It must not include other time interval such as SSL handshake and SOCKS authentication.
IMO the NT spec has the better definition, as secureConnectionEnd==connectEnd in this case (which is why secureConnectionEnd was omitted from both of the specs).  Also, the 'TCP' phase in the images in both NT and RT specs shows connectEnd including SSL/SOCKS.

[cid:image001.png@01CE35CF.FB89AF50]


- Nic

http://nicj.net/

@NicJ
On 4/10/2013 10:45 AM, Austin,Daniel wrote:
There is no such animal as 'SecureConnectionEnd', in either nav or res timing. It's a significant flaw in the model. Also missing are details about the underlying OCSP calls. This significantly reduces the utility of the spec, IMHO.

R,
D-

Sent from my iPhone

On Apr 10, 2013, at 3:56 AM, "James Simonsen" <simonjam@google.com<mailto:simonjam@google.com>> wrote:
I can only guess it's because that's covered by sslConnectStart/End. But in the case of browsers that don't provide that, it seems like they should fall back to including it connectStart/End. Anyone else have an opinion?

James

On Tue, Apr 9, 2013 at 12:16 PM, Andy Davies <dajdavies@gmail.com<mailto:dajdavies@gmail.com>> wrote:
I understand why the spec states that connectEnd excludes SOCKS authentication etc., but don't quite understand why it excludes the SSL Handshake

"connectEnd must include the time interval to establish the transport connection. It must not include other time interval such as SSL handshake and SOCKS authentication."

I've had a hunt back through the archives but I couldn't find any reference as to why.

Is anyone able to explain?

Thanks

Andy