W3C home > Mailing lists > Public > public-html@w3.org > August 2007

Re: IE's AND EVERYONE ELSE'S object implementation problems (was RE: Baby Steps or Backwards Steps?)

From: Doug Schepers <schepers@w3.org>
Date: Tue, 21 Aug 2007 17:25:19 -0400
Message-ID: <46CB583F.8040605@w3.org>
To: Robert Burns <rob@robburns.com>
CC: HTML Working Group <public-html@w3.org>, wai-xtech@w3.org, Sander Tekelenburg <st@isoc.nl>, "Gregory J. Rosmaita" <oedipus@hicom.net>, Chris Wilson <Chris.Wilson@microsoft.com>

Hi, Robert-

Thanks for putting these tests together.  I would also like to see how 
SVG content is handled for all of these cases.  There are a few 
scenarios to consider:

* static SVG (no interactivity, just an image)
* dynamic SVG (animation, audio, and/or video... all of these have 
time-based aspects)
* SVG with various "intrisic size" issues... no width and/or height 
attribute, width/height of 100%, width/height as absolute values, and 
variations involving viewboxes.

Finally, since SVG has its own DOM, the tests should include access to 
and from that DOM.  I can't think of another example of a format that 
would be embedded in an <object> that also has its own DOM, but it could 
come to pass... what should/could happen when HTML is embedded in an 
HTML <object> element?

Obviously, IE won't handle SVG at all out of the box (though I hope that 
will change), but there are plugins available (Adobe's being the most 
famous).  Opera will most likely handle all the cases very handily, and 
FF and Safari will handle some of it (interactivity through links and 
script alone, no native animation, audio, or video).

I'm rather busy at the moment, but if you need help putting together 
test cases, I'm happy to do so (though no promises as to when).  A long 
while back, I did put together some simple tests for the link target 
attribute values that might help get us started:

http://vectoreal.com/w3c/link-target-test/


Regards-
-Doug Schepers
W3C Staff Contact, SVG, CDF, and WebAPI


Robert Burns wrote (on 8/21/2007 10:08 AM):
> 
> Hello all,
> 
> I've intended to put together some tests for some time, and Chris's 
> question prompted me to do it. I find it crazy the steps authors go 
> through to display non-text media into their pages. In my view, OBJECT 
> should should just be a simple container that an author can attach any 
> of the most common media types two to embed content. For all of the 
> common cases, it should be as simple as:
> 
> <object data='afile'>the fallback content</object>
> 
> Well, we may be getting closer to that goal. I put together a page to 
> test the ability of browsers to handle this simplest authoring practice. 
> This page is to test the treatment of the OBJECT element in various 
> browsers and other UAs. It includes images that have intrinsically small 
> dimensions and intrinsically large dimensions repeated three times for: 
> PNG, GIF and JPEG. There are also three time-oriented files: an MP3 
> audio file; a video file using 3rd Generation Partnership Project 3GP 
> container with H.264 content.baseline profile; and a Flash video file 
> (SWF).
> 
> The 3GP might not have been the best choice since I'm not sure how much 
> support there is for that format. I think the other formats should be 
> widely supported in most user's environments. The page is located here:
> 
> <http://lists.w3.org/Archives/Public/www-archive/2007Aug/att-0027/TestObject.html> 
> 
> 
> What I've found is this:
> 
>  • Safari 3.0 beta: Poor support for images beyond particular intrinsic 
> dimensions. It expects images to be smaller than a particular size and 
> then provides scrollers for any larger images. Time-based media (audio 
> and video) do not load,.except  for flash.
>  • Safari 3.0 beta with recent nightly build of WebKit: The large still 
> images problems are fixed. Still no luck with time-based media except 
> for flash.
>  • Mozilla (2.0.0.6): Still images work fine: both small and large.  
> Time-based media does not load except for the audio track, which plays 
> automatically on load.
> • Opera (9.0.2 build3512)  handles still images fine. Time-based media 
> is also handled nicely in a sane default way. All media begins playing 
> immediately on load. Video is visible. Audio is audible. Small 
> unobtrusive controls are provided for: forward, revers, volume, 
> play/pause, and an quicktime menu.
> • iCab (3.0.3)  Same as Opera. Handles still images fine. Time-based 
> media is also handled nicely in a sane default way. All media begins 
> playing immediately on load. Video is visible. Audio is audible. Small 
> unobtrusive controls (probably from QuickTime) are provided for: 
> forward, revers, volume, play/pause, and an quicktime menu.
>  • IE7 (7.0.5730.11): Fails every test. It presents the fallback content 
> for every image, video and audio file (sorry Chris). This could be due 
> to the video formats I chose, but since the images were not handled 
> either I doubt it. I would imagine that anyone with a recent 
> QuickTime/iTunes installed on Windows should be able to handle this 
> content.
> 
> A couple of caveats here. These are all very current browser releases. 
> Who knows how well this simple approach works for the large base of 
> older browsers out there. It looks from these results that OBJECT still 
> could not replace IMG as a simple container for still images providing a 
> fallback mechanism even in the latest releases.
> 
> For video (including flash) and audio Opera is the hands down winner. It 
> basically passed every test. What a simpler World we would live in if 
> every browser handled other media so nicely. I could imagine changing 
> minor things about Opera's implementation, but for the most part, the 
> other browsers just need to do what Opera is doing (for  example, the 
> audio OBJECT would be better handled if the CSS box generated for it 
> wrapped only the controls by default; vertical-alignment and other CSS 
> properties might be used to further change that).
> 
> The one idea that occurred to me in putting together and running these 
> tests is it might be desirable to have an attribute on OBJECT to turn on 
> or off time-based controls. Perhaps this is something CSS should handle, 
> though in these sorts of inherently visual objects, I'm inclined to 
> think it should be handled in both layers: semantic and presentation.
> 
> Take care,
> Rob
> 
> On Aug 21, 2007, at 5:07 AM, Sander Tekelenburg wrote:
> 
>>
>> At 10:01 -0700 UTC, on 2007-08-20, Chris Wilson wrote:
>>
>> [...]
>>
>>> Could someone describe the exact issue, and tell me if it is still 
>>> present
>>> in IE7?  I know we still don't deal with 404 errors properly, but OBJECT
>>> fallback was dramatically improved in IE7.
>>
>> Just search the public-html@w3.org Archive. (There's plenty of specific
>> questions to you/Microsoft that haven't been responded to yet.) Here's a
>> selection:
>>
>> <http://www.w3.org/mid/op.tuheo3gdwxe0ny@widsith.local>
>> <http://www.w3.org/mid/a9699fd20706260653ua3f03d6h4ccbff6ea11ab158@mail.gmail.com> 
>>
>> <http://esw.w3.org/topic/ObjectTestResults>
>> <http://www.w3.org/mid/468A52F8.2000100@Rhul.Ac.Uk>
>> <http://www.w3.org/mid/468AE397.3000200@fastmail.us>
>>
>> And two relevant test cases:
>> <http://santek.no-ip.org/~st/tests/object/>
>> <http://santek.no-ip.org/~st/tests/picturetag/>
>>
>> I don't have IE7 available, but I'm sure you do ;)
> 
> 

-- 
Received on Tuesday, 21 August 2007 21:25:40 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:16:04 GMT