RE: [xml-dev] Beware of XPath expressions that produce falsepositives

["Costello, Roger L." <costello@mitre.org>]

Dimitre wrote:

You haven't defined what should "empty" mean exactly.
               Therefore, *any* XPath expression attempting to express
               this undefined concept of "empty" is potentially wrong.

 

Fantastic Dimitre! You are spot on. I did not define what I mean by “empty element”.

 

I would like to do so now.

 

First, examples of what I consider to be empty elements and examples of elements that I consider not empty.

 

In the following instance document, the B element is empty:

 

<Row>
    <A>foo</A>
    <B/>
    <C>bar</C>
</Row>

 

In the following instance document the B element is empty:

<Row>
    <A>foo</A>
    <B></B>
    <C>bar</C>
</Row>

 

In the following instance document the B element is empty:

<!DOCTYPE Row [
<!ENTITY null ''>
]>
<Row>
    <A>foo</A>
    <B>&null;</B>
    <C>bar</C>
</Row>

 

In the following instance document the B element is not empty:

 

<Row>
    <A>foo</A>
    <B>   </B>
    <C>bar</C>
</Row>

 

In the following instance document the B element is not empty:

 

<Row>
    <A>foo</A>
    <B><!-- Hello, world --></B>
    <C>bar</C>
</Row>

 

In the following instance document the B element is not empty:

 

<Row>
    <A>foo</A>
    <B><?my-pi x="blah"?></B>
    <C>bar</C>
</Row>

 

In the following instance document the B element is not empty:

 

<Row>
    <A>foo</A>
    <B><bad/></B>
    <C>bar</C>
</Row>

 

In the following instance document the B element is not empty:

 

<Row>
    <A>foo</A>
    <B>99</B>
    <C>bar</C>
</Row>

 

In the following instance document the B element is not empty:

 

<Row>
    <A>foo</A>
    <B x="10"/>
    <C>bar</C>
</Row>

 

Those examples illustrate what I mean by an “empty element.” Now for a definition.

 

[Definition] An element is empty if and only if the element consists of a text node containing a string of length zero and nothing else (no attribute, comment, processing instruction, child element, or non-string value).

 

Is that definition 100% precise? Does it cover every possible case? Can it be stated more succinctly and precisely?

 

/Roger