Is this valid XML?

Question

I know that syntax is valid, but my question is whether it is logically valid:

<parent>
    <name>John</name>
    <child>Mary</child>
    <child>Lucy</child>
    <child>Hannah</child>
</parent>

or a proper way to do this is:

<parent>
    <name>John</name>
    <child>
        <name>Mary</name>
    </child>
    <child>
        <name>Lucy</name>
    </child>
    <child>
        <name>Hannah</name>
    </child>
</parent>

Is there some document online that definitely says what's right and wrong?

Answer 1

It depends on what you are going to do with it. The 2nd version is better if there's a chance you will need to store more data about each person in the future.

Answer 2

I prefer the latter, as it makes clear that it's the NAME of the child that is Mary, and not that the CHILD ITSELF that is Mary.

I think that using attributes is even better, like so:

<parent name="John">
    <child name="Mary" />
    <child name="Lucy" />
    <child name="Hannah" />
</parent>

because it makes it clear that the name is just a characteristic of the parent/child entity.

Answer 3

The second is preferable. This makes it clear that name is a property of a child and does not identify the child iteself.

Think of it in terms of classes:

This

class Parent {
    string Name;
    List<Child> Children;
}

class Child {
    string Name;
}

is preferable to

class Parent {
    string Name; 
    List<string> Children;
}

The second option also gives you the flexibility to expand in the future (add a birthday element, for example).

The more subjective debate is whether to use elements or attributes for properties like name, etc.

Finally, add a children element with the child elements contained there.

Answer 4

The second one seems to make more sense from an extensibility point of view. What happens if you need to add birthday for a child in the first one? Yes, you could add an XML attribute but you'll end up sooner or later getting stuck on adding a complex type or even basic enum to it.

Also - it may be better to group the child elements under a single 'children' element:

<parent>
    <name>John</name>
    <children>
      <child>
        <name>Mary</name>
        <dob>1970-01-01</dob>
      </child>
      <child>
        <name>Lucy</name>
        <dob>1971-01-01</dob>
      </child>
      <child>
        <name>Hannah</name>
        <dob>1974-01-01</dob>
      </child>
    <children>
</parent>

One more thing: you probably wouldn't group the children under one single parent element, but I've left that in-line with your original.

Answer 5

It is XML, there is no right or wrong. Both your answers are correct, however this is equally valid:

<parent>
  <name>John</name>
  <children>
    <child>Mary</child>
    <child>Lucy</child>
    <child>Hannah</child>
  </children>
</parent>

Which way should you choose? It depends on the task. I do not think any of the ways presented are the most flexible (what if the children have children?)

Answer 6

There is no right answer to this question, of course, but if the child is more complex (or could grow to be more complex) than a single string of text, then the second option is preferable.

In terms of what you usually see, generally in either case all the child elements would be grouped under a children element. In certain visualization environments, it can help to just close away all the children while other elements retain the focus.

Answer 7

Both are right. There are no definition on how to structure your element - it's entirely up to you!

Some people tries to minimice the number of nodes. And those people would maybe create the xml like

<parent name="John">
    <child name="Mary" />
    <child name="Lucy" />
    <child name="Hannah" />
</parent>

But the clue about XML is that you should always make it as easly as possible to read and understand, for human beeings. Screw the comupters, they will always understand you XML, so make it human readble!

Answer 8

I would use an alternative between Shoko and DanDan or Wim Hollebrandse:

<parent name="John">
  <children>
    <child name="Mary" />
    <child name="Lucy" />
    <child name="Hannah" />
  </children>
</parent>

because I like the "set" of child which are actually children.

Answer 9

There are no standards for how to map data to an XML Schema. There are some common practices, one of which is to use striped XML, so nested element take type/relation/type/relation roles alternately:

<!-- striped style, RDF etc -->
<person>
    <name>John</name>
    <children>
        <person>
            <name>Mary</name>
        </person>
        <person>
            <name>Lucy</name>
        </person>
        <person>
            <name>Hannah</name>
        </person>
    <children>
</parent>

This is very regular, but somewhat more verbose.

It's generally a bad idea to put human readable text into attributes to save space:

<person name="fred"/>

As that precludes use of ruby mark up which is necessary for some forms of internationalisation, as well as being more complicated to render using CSS. If you're only concerned with compact representation and ASCII text, XML might not be the best format to be working with.

Answer 10

It's worth mentioning that the term "valid" has a specific meaning in XML.

An XML document is valid if and only if it conforms to its DTD or schema. Basically, the universe of strings of text is divided into two categories: those that are well-formed XML, and those that aren't. The universe of well-formed XML documents also is divided into three categories: valid XML documents (which conform to their DTD/schema), invalid XML documents (which don't), and those whose validity cannot be determined (because they don't have a DTD/schema).

As far as your actual question goes, you can only judge the design of an XML document on the basis of its fitness to the purpose for which it is to be used. Are you going to be transforming it with XSLT? Querying it with XPath? Processing it with Linq-to-XML? Processing it with a SAX reader? Deserializing the data in it into objects? Editing it in Notepad? Validating it against a schema? Transporting it over a slow network? All of those things (and there are many more) should influence the design of your XML. There is no one right answer.