The second way isn't valid XML; did you mean <numbers>[3,2,1]</numbers>?
If so, then the first one is preferred because all you need to get the array elements is some XML manipulation. On the second one you first need to get the value of the <numbers> element via XML manipulation, then somehow parse the [3,2,1] text using something else.
Or if you really want some compact format, you can consider using JSON (which "natively" supports arrays). But that depends on your application requirements.
As its name is "numbers" it is clear it is a list of number... So an array of number... no need of the attribute type... Although I like the principle of specifying the type of field in a type attribute...
Once I've seen such an interesting construction:
<Ids xmlns:id="http://schemas.microsoft.com/2003/10/Serialization/Arrays">
<id:int>1787</id:int>
</Ids>
In XML values in text() nodes.
If we write this
<numbers>1,2,3</numbers>
in element "numbers" will be one text() node with value "1,2,3".
Native way to get many text() nodes in element is insert nodes of other types in text.
Other available types is element or comment() node.
Split with element node:
<numbers>3<_/>2<_/>1</numbers>
Split with comment() node:
<numbers>3<!---->2<!---->1</numbers>
We can select this values by this XPath
//numbers/text()
Select value by index
//numbers/text()[3]
Will return text() node with value "1"
No, there is no simpler way. You only can lose the type=array.
<numbers>
<value>3</value>
<value>2</value>
<value>1</value>
</numbers>
Source: Stackoverflow.com