W3C home > Mailing lists > Public > public-web-perf@w3.org > May 2013

Re: Updates to Navigation Timing errata

From: Boris Zbarsky <bzbarsky@MIT.EDU>
Date: Tue, 14 May 2013 13:27:40 -0400
Message-ID: <5192740C.7040108@mit.edu>
To: "Austin,Daniel" <daaustin@paypal-inc.com>
CC: "public-web-perf@w3.org" <public-web-perf@w3.org>
On 5/14/13 1:13 PM, Austin,Daniel wrote:
> Hmmm...while this may be true for the overall page composition/loading process, for any given resource, the steps will occur in this deterministic order, with a DNS resolution step preceding an attempt to establish a TCP, etc.

No.  My point is that they will NOT occur in this order in actual practice.

> It's safe to say that a DNS resolution step will always precede the TCP step etc, even if these steps are performed out of order or speculatively.

In what sense is it safe to say that, then?

> Even 'speculative preconnects' require a DNS resolution step (which may be local, from the system cache or even /etc/hosts).

Well, yes, but the point is that if the local cache entry has expired 
you might start a TCP connection to the IP anyway and in parallel do a 
DNS lookup; when the DNS lookup completes and (common case) confirms 
that the IP has not changed your TCP connect may be all done already and 
you're ready to go.

> If the UA performs speculative lookups for objects not retrieved, say on mouseover(), do these time segments show up in our results, even if the object is never requested?

The spec doesn't say, since it doesn't even consider this situation.

-Boris
Received on Tuesday, 14 May 2013 17:28:09 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:04:35 UTC