Re: [server-timing] Linking to a ResourceTiming entry

Yep, this smells like another instance of same problem described here:
https://lists.w3.org/Archives/Public/public-web-perf/2015Feb/0062.html

On Wed, Mar 4, 2015 at 8:46 AM, Todd Reifsteck <toddreif@microsoft.com>
wrote:

>  Ilya previously suggested a resourceId uses to associate resource
> timings for identifying relationships such as preflight->xhr. I generally
> agree that it could be re-used on ServerTiming for association.
>
>
>
> What are other folk’s thoughts?
>
>
>
> -Todd
>
>
>
> *From:* Nic Jansma [mailto:nic@nicj.net]
> *Sent:* Wednesday, March 4, 2015 7:52 AM
> *To:* public-web-perf@w3.org
> *Subject:* [server-timing] Linking to a ResourceTiming entry
>
>
>
> Hi everyone,
>
> From my understanding of ServerTiming, you'll get an entry in the
> PerformanceTimeline with these attributes:
>
> name (URL of resource)
>
> entryType, startTime, duration (common PerformanceTimline attributes)
>
> metric
>
> description
>
>  The example shows how you can get the ServerTiming (and ResourceTiming)
> for a particular resource:
>
>  var serverMetrics = window.performance.getEntriesByName('https://example.com/resource.jpg');
>
> for (i = 0; i < serverMetrics.length; i++) <
>
> /span>{
>
>     entry = serverMetrics[i];
>
>     if (entry == "server") {
>
>       // entry.metric, entry.duration, entry.description
>
>     }
>
> }
>
> But what if the same resource is downloaded twice for some reason?  For
> example, if there are non-cachable headers on an XHR.  How can you be sure
> which ServerTIming goes with which ResourceTiming?
>
> Is the startTime guaranteed to be the same for the ServerTiming and
> ResourceTiming entries?  That would be one way to "link" them.
>
> If not, should we add an additional attribute to link the two in some
> way?  A "resourceId" that is the same on both ServerTiming and
> ResourceTiming entries?
>
> It's a bit of an edge case, but something that would be helpful to think
> through now.
>
>  --
>
> - Nic
>
>