What does "xmlns" in XML mean?

It means XML namespace.

Basically, every element (or attribute) in XML belongs to a namespace, a way of "qualifying" the name of the element.

Imagine you and I both invent our own XML. You invent XML to describe people, I invent mine to describe cities. Both of us include an element called name. Yours refers to the person’s name, and mine to the city name—OK, it’s a little bit contrived.

<person>
    <name>Rob</name>
    <age>37</age>
    <homecity>
        <name>London</name>
        <lat>123.000</lat>
        <long>0.00</long>
    </homecity>
</person>

If our two XMLs were combined into a single document, how would we tell the two names apart? As you can see above, there are two name elements, but they both have different meanings.

The answer is that you and I would both assign a namespace to our XML, which we would make unique:

<personxml:person xmlns:personxml="http://www.your.example.com/xml/person"
                  xmlns:cityxml="http://www.my.example.com/xml/cities">
    <personxml:name>Rob</personxml:name>
    <personxml:age>37</personxml:age>
    <cityxml:homecity>
        <cityxml:name>London</cityxml:name>
        <cityxml:lat>123.000</cityxml:lat>
        <cityxml:long>0.00</cityxml:long>
    </cityxml:homecity>
</personxml:person>

Now we’ve fully qualified our XML, there is no ambiguity as to what each name element means. All of the tags that start with personxml: are tags belonging to your XML, all the ones that start with cityxml: are mine.

There are a few points to note:

If you exclude any namespace declarations, things are considered to be in the default namespace.

If you declare a namespace without the identifier, that is, xmlns="http://somenamespace", rather than xmlns:rob="somenamespace", it specifies the default namespace for the document.

The actual namespace itself, often a IRI, is of no real consequence. It should be unique, so people tend to choose a IRI/URI that they own, but it has no greater meaning than that. Sometimes people will place the schema (definition) for the XML at the specified IRI, but that is a convention of some people only.

The prefix is of no consequence either. The only thing that matters is what namespace the prefix is defined as. Several tags beginning with different prefixes, all of which map to the same namespace are considered to be the same. For instance, if the prefixes personxml and mycityxml both mapped to the same namespace (as in the snippet below), then it wouldn't matter if you prefixed a given element with personxml or mycityxml, they'd both be treated as the same thing by an XML parser. The point is that an XML parser doesn't care what you've chosen as the prefix, only the namespace it maps too. The prefix is just an indirection pointing to the namespace.

Attributes can be qualified but are generally not. They also do not inherit their namespace from the element they are on, as opposed to elements (see below).

Also, element namespaces are inherited from the parent element. In other words I could equally have written the above XML as

<person xmlns="http://www.your.example.com/xml/person">
    <name>Rob</name>
    <age>37</age>
    <homecity xmlns="http://www.my.example.com/xml/cities">
        <name>London</name>
        <lat>123.000</lat>
        <long>0.00</long>
    </homecity>
</person>

---

It defines an XML Namespace.

In your example, the Namespace Prefix is "android" and the Namespace URI is "http://schemas.android.com/apk/res/android"

In the document, you see elements like: <android:foo />

Think of the namespace prefix as a variable with a short name alias for the full namespace URI. It is the equivalent of writing <http://schemas.android.com/apk/res/android:foo /> with regards to what it "means" when an XML parser reads the document.

NOTE: You cannot actually use the full namespace URI in place of the namespace prefix in an XML instance document.

Check out this tutorial on namespaces: http://www.sitepoint.com/xml-namespaces-explained/

---

I think the biggest confusion is that xml namespace is pointing to some kind of URL that doesn't have any information. But the truth is that the person who invented below namespace:

xmlns:android="http://schemas.android.com/apk/res/android"

could also call it like that:

xmlns:android="asjkl;fhgaslifujhaslkfjhliuqwhrqwjlrknqwljk.rho;il"

This is just a unique identifier. However it is established that you should put there URL that is unique and can potentially point to the specification of used tags/attributes in that namespace. It's not required tho.

Why it should be unique? Because namespaces purpose is to have them unique so the attribute for example called background from your namespace can be distinguished from the background from another namespace.

Because of that uniqueness you do not need to worry that if you create your custom attribute you gonna have name collision.

---

xmlns - xml namespace. It's just a method to avoid element name conflicts. For example:

<config xmlns:rnc="URI1" xmlns:bsc="URI2">
  <rnc:node>
      <rnc:rncId>5</rnc:rncId>
  </rnc:node>

  <bsc:node>
      <bsc:cId>5</bsc:cId>
  </bsc:node>
</config>

Two different node elements in one xml file. Without namespaces this file would not be valid.

---

You have name spaces so you can have globally unique elements. However, 99% of the time this doesn't really matter, but when you put it in the perspective of The Semantic Web, it starts to become important.

For example, you could make an XML mash-up of different schemes just by using the appropriate xmlns. For example, mash up friend of a friend with vCard, etc.

---