Re: HTML imports and rendering

Thanks Morrita-san and Dimitri for clarifying, that sounds great.

Also to be sure, just like styles, could HTML imports be injected after the
page load to dynamically load? Also would the `onload` event be triggered
accordingly as well on completion for this case?

Guy

On 24 June 2013 11:19, Dimitri Glazkov <dglazkov@chromium.org> wrote:

> Morrita-san is right.
>
> Guy, the only time you would see FOUC with imports is in the same
> situations when you see FOUC with styles: when the rendering engine
> decided to give up the import ever loading.
>
> :DG<
>
> On Sun, Jun 23, 2013 at 11:31 PM, Hajime Morrita <morrita@google.com>
> wrote:
> > Hello,
> >
> > HTML Imports spec relies on the concept of "things that is blocking
> > scripts".
> > In HTML, <link>-ed styles are blocking scripts. So All styles that
> precede a
> > <script> are loaded before the script execution.
> >
> http://www.whatwg.org/specs/web-apps/current-work/multipage/semantics.html#the-link-element
> >
> > HTML Imports behaves in similar manner. The <link>s to HTML Import block
> > following <script> execution.
> > So in practice, linked imports are loaded before the page rendering like
> > liked styles are applied before that,
> > because most of the pages will have some kind of initialization script.
> >
> >
> > Does this make sense?
> >
> > # @dglazkov correct me if I'm wrong.
> >
> > --
> > morrita
> >
> >
> >
> > On Mon, Jun 24, 2013 at 6:45 AM, Guy Bedford <guybedford@googlemail.com>
> > wrote:
> >>
> >> HTML imports don't seem to block the rendering of the HTML page at all.
> >>
> >> If this is the case, custom elements must always be progressively
> enhanced
> >> with a flash of content and style. This is not always ideal. It can make
> >> sense for certain custom elements to only display after having their
> style
> >> and initial scripts run.
> >>
> >> Has there been any discussion around having attributes on the link tag
> to
> >> specify blocking behaviour at all? Surely it is important to have
> control
> >> over HTML rendering at this level?
> >>
> >> Yes, build tools / multiplexing should occur for these blocking parts
> when
> >> used in production for performance reasons, but I don't believe the fact
> >> that it can become a performance issue is reason enough to force all
> >> components to use progressive enhancement. Or am I missing something
> here?
> >
> >
> >
> >
> > --
> > morrita
>