Pros/Cons for containers for child nodes in XML?

Question

I've noticed that different XML schemas define child elements differently. Some define them directly under the parent nodes like so:

<parent>
    <foo />
    <foo />
    ...
    <foo />
    <bar />
    <bar />
    ...
    <bar />
</parent>

where as others define container nodes around the child nodes like so:

<parent>
    <foos>
        <foo />
        <foo />
        ...
        <foo />
    </foos>
    <bars>
        <bar />
        <bar />
        ...
        <bar />
    </bars>
</parent>

I haven't had any issues serializing/deserializing into either format as necessary, and I can't think of any reasons to prefer one over the other.

What (if any) are the pros/cons of each approach?

Answer 1

Well, the second one is more "intuitive". Is more hierarchical. For example you have this one:

<animal>
 <snakes>
   <boa />
   <python />
 </snakes>
 <monkeys>
   <red_ass_monkey />
   <yellow_monkey />
 </monkeys>
</animal>

I would definitely choose the second one when your objects are somehow hirearchical linked, because is more logical. Also when you read yourself to modify something directly in the xml (just in case) you will find your way better, knowing that a python is a snake which is an animal (This one when thing get complicated), better than searching in the whole list.

Answer 2

one possibility for such a structure would be if you needed to group some <foo/>s together into a repeating set:

`

<foos id=1>
    <foo />
    <foo />
    ...
    <foo />
</foos>
<foos id=2>
    <foo />
    <foo />
    ...
    <foo />
</foos>

`

Answer 3

The first scheme is extensible and not that hard to implement, you just iterate over all childNodes of parent and look whether namespace and element name match anything you can read. However, sometimes, splitting can be favorable, especially when the processing the bars depends on all foos or so.

To borrow an example:

<zoo xmlns="http://example.org/zoo" xmlns:z="http://example.org/zoo"&gt;
<cages>
  <cage name="open-air" />
  <cage name="glass-cage" />
</cages>
<animals>
  <monkey name="Orlan" cage="open-air"/>
  <monkey name="Jeremey" cage="glass-cage"/>
  <snake name="spssshs" cage="glass-cage"/>
  <panda xmlns="http://china.cn/zoo" z:name="Ying Ying" z:cage="open-air"/>
</animals>
</zoo>

So, separating cages and animals makes sense. However, if you had grouped the animals in monkeys and snakes, you would need to add lots of extra processing logic for pandas.