W3C home > Mailing lists > Public > public-tt@w3.org > February 2009

RE: semantics of unbound region attribute.

From: Sean Hayes <Sean.Hayes@microsoft.com>
Date: Wed, 11 Feb 2009 21:11:48 +0000
To: "Glenn A. Adams" <gadams@xfsi.com>, Public TTWG List <public-tt@w3.org>
Message-ID: <AB3FC8E280628440B366A29DABB6B6E802EE4C1D56@EA-EXMSG-C334.europe.corp.microsoft.com>
9.3.1 only applies when no region is supplied. My issue is when there is more than one region supplied.

Consider my example:

   <div region="r1">
     <p tts:color="red" begin="0s" end="10s">This text must be red.</p>
     <div region="r2">
         <p tts:color="blue" begin="0s" end="10s">This text must be blue.</p>

My expectation as an author is that the first p is going to show up in region r1, and the second in r2 (assuming these are defined, thus no implied region semantics are applying here)

However if we apply the algorithm in 3.4.2

 1.  for each temporally active region R, replicate the portion of the document's sub-tree headed by the body element;

so we have 2 copies, one for each region

2.    evaluating this sub-tree in a postorder traversal, prune elements if they are not a content element, if they are temporally inactive, if they are empty, or if they aren't associated with region R;

When we reach the <body> element in each traversal we have a problem, as this is not associated with any region:

n  The element does not supply a region attribute.

n  No ancestor of the element specifies a region attribute.

n  No default region is implied.

Thus the <body> gets pruned.

If we cannot specify <body region="r1 r2">, then how are we to associate the body with both regions so it doesn't get pruned?

I think we either need to allow IDREFS for the region attribute, or need some other language that applies when no region attribute is defined, but default region processing is not happening. Such as:

1.    if a default region was implied (due to the absence of any region<http://dev.w3.org/2008/tt/spec/ttaf1-dfxp.html#layout-vocabulary-region> element), then the element is associated with the default region.  Otherwise, if the element has a non empty subtree, the element is associated with the region R of the intermediate tree defined above.


From: Glenn A. Adams [mailto:gadams@xfsi.com]
Sent: Friday, January 30, 2009 9:00 AM
To: Sean Hayes; Public TTWG List
Subject: Re: semantics of unbound region attribute.

Actually, the text in 9.3.1 effectively defines the algorithm for associating content elements with regions. See the REGION-OF() pseudo function. However, since this was an example of an implementation of the (prose) description of the algorithm just above the example, what we need to do is add normative text that defines this association. I have done that in the just posted draft, and marked Issue #42 as closed.

Note that the example you give below of:

<body region="r1 r2">

is not presently supported, since the value of the region attribute must be a single IDREF.


On 12/11/08 1:20 AM, "Sean Hayes" <Sean.Hayes@microsoft.com> wrote:
Another issue that has concerned me lately is what it means when the region attribute is unspecified. According to the spec, this simply means the sub-tree does not show up in the rendering, Which is fine and simple to understand and implement; however I'm wondering if it should perhaps instead have the special meaning of 'any region'.

For example:
   <div region="r1">
     <p tts:color="red" begin="0s" end="10s">This text must be red.</p>
     <div region="r2">
         <p tts:color="blue" begin="0s" end="10s">This text must be blue.</p>

In this case, clearly the intent is that div 1 shows up in r1 and div 2 shows up in r2, however without specifying <body region="r1 r2">  this won't happen. In complex scenarios I can imagine the amount of region attributes getting quite tedious.

Also, if as we have discussed, we are going to allow the idea of an anonymous default region, which I think most of the test suite now relies on, we will need some sort of rule like this anyway in order to get content to be targeted at that anonymous region.

We could still allow that an explicit declaration of no region <body region=""> means that it will show up in no region at all to retain the current semantics.

Sean Hayes
Media Accessibility Strategist
Accessibility Business Unit

Office:  +44 118 909 5867,
Mobile: +44 7875 091385
Received on Wednesday, 11 February 2009 21:11:09 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 5 October 2017 18:24:04 UTC