Re: media fragment URI use on web pages

On Thu, 28 Oct 2010 02:13:07 +0200, Silvia Pfeiffer  
<silviapfeiffer1@gmail.com> wrote:

> Hi guys,
>
> I've been wondering about how we can use media fragment URIs on Web
> page URLs such that the fragment is handed through to the correct
> media element. Existing schemes - such as YouTube's scheme of e.g.
> http://www.youtube.com/watch?v=ZhMoxXilwro#t=186 only work with a
> single video on a page.
>
> It might be an idea to suggest something like:
> http://example.com/page.html#video[0]&t=10,20
>
> Then it's possible to provide a URL with media fragments for multiple
> videos on a page:
> http://example.com/page.html#video[0]&t=10,20&video[1]&t=30,40 etc.
>
> or for all videos on the page:
> http://example.com/page.html#videos&t=10,20
>
> It would be nice if something like this (or nicer - improved
> suggestions welcome) becomes a scheme that everyone uses and that
> therefore the browsers can support.
>
> It's a scheme on a Web page (.html) rather than on a media resource
> (.ogv / .webm / .mp4) and as such not really something that this group
> was chartered for. But I believe we could add a note that recommends
> such use and would be a Web author recommendation, and that the HTML
> WG could eventually pick up as a browser recommendation.

While this would be very useful, it's not something that can be  
standardized in browsers. The URL page.html#t=1 already causes browsers to  
scroll to the element with id="t=1". Overloading this behavior would most  
likely break some pages. See  
<http://www.whatwg.org/specs/web-apps/current-work/multipage/history.html#scroll-to-fragid>

This is quite unfortunate, as far as I can see the best we can hope for is  
page.html?t=1 with per-site server-side solutions or page.html#t=1 with  
per-site JavaScript solutions (those sites would have to make sure to not  
have id="t=1" on their pages, and live with not being able to use the  
fragment for its usual purpose). Perhaps there's room for standardization  
here, as long as it's clear that User Agents aren't involved.

-- 
Philip Jägenstedt
Core Developer
Opera Software

Received on Thursday, 4 November 2010 10:20:59 UTC