Re: [minutes] Web Performance WG Teleconference #122 2013-12-11

I've updated NavigationErrorLogging to explicitly note that the browser
should not retry reporting the error.

Arvind


On Wed, Dec 11, 2013 at 8:09 PM, Ilya Grigorik <igrigorik@google.com> wrote:

>
> On Wed, Dec 11, 2013 at 5:54 PM, Jatinder Mann <jmann@microsoft.com>wrote:
>
>> Ilya also raised a concern that postpone should be able to be set on
>> container elements that aren’t visible, like the div element. This way
>> users can set postpone on the container instead of setting it on each
>> individual element. Seeing that using CSS won’t work as formatting occurs
>> much later in the pipeline, this seems like a reasonable request.
>
>
> To provide a bit more context, the case I had in mind is something like
> this:
>
> <div id="header"> ... </div>
> <div id="hero-content" *postpone*>
>    <img src="large-image.jpg">
>    <script src="awesome-widget.js">
> </div>
> ...
> <my-awesome-future-component *postpone*>
>   <li> Foo
>   <li> Bar
> </my-awesome-future-component>
> </div id="footer"> ... </div>
>
> When designing an RWD site it's fairly common to hide (display: none) an
> entire section (e.g. the big "hero" section in example above), which may
> contain images, scripts, and other external resources. Today, you would
> have to put a "postpone" attribute on every element within that container,
> and even with that you wouldn't be able to postpone a script or any other
> element that's not part of the render tree.
>
> Instead, I'm wondering if we can allow postpone to be defined not just on
> an individual element but on a containing element, which would then
> propagate this state to all of its children - i.e. in example above,
> neither large-image.jpg or awesome-widget.js would get downloaded while
> #hero-content is not part of the rendering tree. I think this would make
> RWD development a lot simpler... and lighter.
>
> As a case in point, I've seen many sites which hide maps script or social
> widgets sections until they are clicked and expanded (and often they're
> not) - today, those scripts are still downloaded, whereas allowing postpone
> on an outer container would load them on demand or not at all.
>
> ig
>
> P.S. we should also think about how this could/would interact with custom
> elements.
>
>