Is it a good practice to have XML document with no-namespace elements in the middle

Question

I have an XML looking like this:

<?xml version="1.0" encoding="utf-8"?>
<n0:OrderConfirmation xmlns:n0="http://company.org/interface/MVSI"
 xmlns:ord="urn:company.com:Order:XYZ:001">
 <MessageHeader>
 <ID>001</ID>
 <CreationDateTime>2020年09月10日T07:39:26.697Z</CreationDateTime>
 </MessageHeader>
 <Order>
 <OrderID>00027847</SalesOrderID>
 <Status>Confirmed</Status>
 <Items>
 <Article>
 <ArticleID>ABC001</ArticleID>
 <LineItem>000001</LineItem>
 </Article>
 </Items>
 </Order>
</n0:OrderConfirmation>

I would say a better way would be to declare a default namespace like this:

<?xml version="1.0" encoding="utf-8"?>
<n0:OrderConfirmation xmlns:n0="http://company.org/interface/MVSI"
 xmlns="urn:company.com:Order:XYZ:001">
 <MessageHeader>
 <ID>001</ID>
 <CreationDateTime>2020年09月10日T07:39:26.697Z</CreationDateTime>
 </MessageHeader>
 <Order>
 <OrderID>00027847</SalesOrderID>
 <Status>Confirmed</Status>
 <Items>
 <Article>
 <ArticleID>ABC001</ArticleID>
 <LineItem>000001</LineItem>
 </Article>
 </Items>
 </Order>
</n0:OrderConfirmation>

Or more simplified: the root element could be part of the default namespace:

<?xml version="1.0" encoding="utf-8"?>
<OrderConfirmation xmlns="urn:company.com:Order:XYZ:001">
 <MessageHeader>
 <ID>001</ID>
 <CreationDateTime>2020年09月10日T07:39:26.697Z</CreationDateTime>
 </MessageHeader>
 <Order>
 <OrderID>00027847</SalesOrderID>
 <Status>Confirmed</Status>
 <Items>
 <Article>
 <ArticleID>ABC001</ArticleID>
 <LineItem>000001</LineItem>
 </Article>
 </Items>
 </Order>
<OrderConfirmation>

because the n0 namespace value is global to the system and for all XML documents generated by the system.

I am wondering if this is a good practice in XML to let the middle elements in the "no-namespace" and alternatively to declare a namespace that is never used.

UPDATE: To give a better context of the situation, we are integrating different systems using an ESB in the whole enterprise. Systems can be in several departments across the company. In the actual situation we are transforming the incoming XML to a more normalized XML with a unique namespace related to the kind of data we're managing (an order in that case), like a canonical data model all other systems have to know about.

Answer 1

Some people think everything should be in a namespace, because it documents who is responsible for the design of the vocabulary. If you follow that line of thought, then this applies to the payload of your example messages as much as it applies to everything else.

Some people think it's a lot simpler to not use namespaces unless you need them.

Depends whether you like a big-project discipline and control approach or a small-project just-get-it-done-quickly approach.

• рüффп: Yes I understand the way of no-namespace concept for simplification, but the example in my question is a mix between both ways: root element is in a namespace while sub-elements are in the no-namespace. If we decide to use no-namespace, I guess this should be globally decided no?
• Michael Kay: I'm not a great believer in "rules of good practice" for this kind of thing. Use the technology in whatever way best meets your project requirements (which you haven't told us anything about).
• рüффп: In my case it's not just a small tutorial about XML but we are ingrating different systems across a whole enterprise (updated the question with context). I think using namespaces is totally recommended in that case.
• Mike Robinson: Namespaces can be very beneficial, especially when you are dealing with many different systems. They greatly improve the ability to validate the XML to be sure that every message that is passed conforms to the expected or prescribed structure.
• Michael Kay: Yes: useful in big complex projects, an unnecessary complication in small simple ones.