Request: 
------

In the part of the spec quoted below, replace the following

      ]] If the image is the child of a figure element that has a child figcaption element [[ 

with this (adding the word "only"):

     ]]  If the image is the _only_ child of a figure element  that has a child figcaption element [[

Spec quote:
--------

]]
If the src attribute is set and the alt attribute is not
    [ snip ]
If the image is the child of a figure element that has a child figcaption element, then the contents of the first such figcaption element are the caption information; abort these steps.
[[

Justification
---------

The above descrption in the spec as it stands, fits for both of these examples:

(1) <figure><figcaption>Foo</figcaption><img src=bar></figure>
(2) <figure><figcaption>Foo</figcaption>Lorem [snipped a long text] ipsum: 
                                                                                               <img src=bar></figure>

However, I believed, that the purpose of  the above step in the algorithm, was to describe what should happen when *exactly* the condidtion shown in the <figure> example in the "Images whose contents are not known" sections is met: http://dev.w3.org/html5/spec/embedded-content-1.html#unknown-images

There we found the following example:

<figure>
 <img src="1100670787_6a7c664aef.jpg">
 <figcaption>Bubbles traveled everywhere with us.</figcaption>
</figure>

In this example, there is no adjacent text close to the <img>, and thus I agree that it makes sense to let the user agent present the <figcaption> as the label.  

However, whenever there is text neary by the <img>, then it does not make sene to to use the <figcaption> as caption for the <img>.