Re: [ResourceTiming] Some clarifications

On Sat, Jun 2, 2012 at 10:57 AM, Jatinder Mann <jmann@microsoft.com> wrote:

>  While looking at some other specs, I had a few questions about Resource
> Timing.****
>
> ** **
>
> Timing-Allow-Origin****
>
> Section 4.5 Cross-origin Resources of the spec [1] defines the
> “Timing-Allow-Origin” HTTP response header and the expected behavior when
> it is present. However, the rules for performing checks with respect to the
> header are not defined. As this portion of the spec was inspired by the
> CORS design [2], should we also include the rules for performance checks
> algorithm as well [3]? I recall when we last discussed this, there was
> hesitation to take a dependency on a spec like CORS and to rather define
> this behavior directly in Resource Timing.
>

Personally, I'd rather rely on someone else's definition than crafting our
own. But if CORS isn't stable enough, then I guess we'll have to roll our
own.


> ****
>
> Web Workers
>
> A few questions arise when thinking of Resource Timing from a Web Workers
> perspective [4]. The Performance interface currently doesn’t implement
> WorkersUtil, meaning this interface isn’t available to web workers. If a
> web worker initiates a download, that data currently will not be reported
> anywhere. Do we want to bring this in scope or out of scope of Resource
> Timing?****
>
> ** **
>
> If we do bring it in, we need clarification on a few things. If a web
> worker initiates a download of a resource, in which document’s timeline
> view should that resource be present? I would suggest that the browsing
> context that created the worker owns the worker and hence the resource
> should show up in that timeline. E.g., Parent A has iframe B and iframe B
> creates a web worker that Parent A also uses. Any resources downloaded by
> the worker will show up in iframe B’s timeline. Thoughts?****
>
> ** **
>
> I think things get a bit more complicated when we talk about shared
> workers that can be shared across top level browsing contexts [5]. If a
> shared worker initiates a download, under which timeline is that download
> shown? I believe a shared worker is its own browsing context. I’m not sure
> what how we would represent that data from a resource timing point of view.
> ****
>
>
I haven't thought about Web Workers at all. Presumably, we'll come across
similar issues with User Timing. Maybe we should just try to ship what
we've got and punt workers out to v2?

I do agree with your points though. I think those make the most sense.

James


> [1]
> https://dvcs.w3.org/hg/webperf/raw-file/tip/specs/ResourceTiming/Overview.html#cross-origin-resources
>
> **
>
> [2] http://www.w3.org/TR/access-control ****
>
> [3] http://www.w3.org/TR/access-control/#resource-sharing-check-0 ****
>
> [4] http://www.w3.org/TR/workers/ ****
>
> [5] http://www.w3.org/TR/workers/#shared-workers-introduction ****
>