[whatwg] whatwg Digest, Vol 81, Issue 25

[Charles Pritchard]

> Date: Thu, 09 Dec 2010 09:24:25 -0500 From: Boris Zbarsky 
> <bzbarsky at MIT.EDU> To: Markus Ernst <derernst at gmx.ch> Cc: 
> whatwg at lists.whatwg.org Subject: Re: [whatwg] Proposal for a tab 
> visibility API Message-ID: <4D00E699.7040909 at mit.edu> Content-Type: 
> text/plain; charset=UTF-8; format=flowed On 12/9/10 3:19 AM, Markus 
> Ernst wrote:
>> >  Am 09.12.2010 00:12 schrieb Boris Zbarsky:
>>> >>  On 12/8/10 5:29 PM, Markus Ernst wrote:
>>>> >>>  Thus, I'd consider an api for detecting the visibility state of every
>>>> >>>  HTML element useful (totally visible, partially visible, hidden - or a
>>>> >>>  percentage value).
>>> >>
>>> >>  This is pretty hard to implement, in general. For example, if I put
>>> >>  another window over the browser window then whether the content in the
>>> >>  browser is visible depends on the exact app running in that window, and
>>> >>  on the parts of it overlapping the browser content, right?
>> >
>> >  Sure. Some applications might even be partially transparent, like some
>> >  fancy-skinned media players.
> Yes, that's precisely what I was saying above.  And it's not just
> "fancy-skinned media players".  It's every single app on Windows 7 by
> default (see "Aero glass").  It's a commonly tweaked setting for
> Terminal on Mac.  It's a common setting for terminal windows on Linux.
>
> The fact is, at this point_most_  windows are partially transparent.
>
> -Boris

Regarding the original proposal, adding "background" to "view-mode" 
seems appropriate:
http://dev.w3.org/2006/waf/widgets-vmmf/

Things considered, even a hidden "tabbed" application is likely 
'minimized', as the browser chrome may be able to take a screen shot of 
its contents.
That said, "background" seems like a reasonable addition to 5.1 View Modes

As for this other idea, of html elements: it's not practical in 
relationship to other windows / applications, as Boris has outlined.

It's somewhat feasible in relationship to the DOM,
querying whether the boundaries of an element extend off-screen. Still, 
I think it would be a computational expensive process.

An example would be, detecting how much of the total body area is 
covered by other elements. I don't see a great use case there;
and I think it could be scripted in current environments by traversing 
through the document, looking at css styles for display, visibility and 
opacity.

Such work isn't really necessary, but for special cases, so I think it's 
better left up to the author to program the functionality if it's needed.



-Charles