Why those two xml string aren't similar

Question

I have two similar xml string. I use XMLUnit to compare them but after I run some sample test to check them it says that they aren't similar and identical. I agree that they aren't identical but I think it should return true for similar. Below are my strings and test code that I run.

<Errors>
  <Error>
    <Sheet>Sample1</Sheet>
    <ErrorCode>4</ErrorCode>
    <Columns>
      <Column>Id</Column>
      <Column>Name</Column>
    </Columns>
  </Error>
  <Error>
    <Sheet>Sample2</Sheet>
    <ErrorCode>4</ErrorCode>
    <Columns>
      <Column>Id</Column>
      <Column>Name</Column>
    </Columns>
  </Error>
</Errors>

and

<Errors>
  <Error>
    <Sheet>Sample1</Sheet>
    <ErrorCode>4</ErrorCode>
    <Columns>  
      <Column>Name</Column>
      <Column>Id</Column>
    </Columns>
  </Error>
  <Error>
    <Sheet>Sample2</Sheet>
    <ErrorCode>4</ErrorCode>
    <Columns>
      <Column>Name</Column>
      <Column>Id</Column>
    </Columns>
  </Error>
</Errors>

The only difference is that Column nodes are reversed but i think it should return that both string are similar.

public void test() throws Exception{
    String myControlXML = "here goes xml1";
    String myTestXML = "here goes xml2";
    Diff myDiff = new Diff(myControlXML, myTestXML);

    System.out.println("pieces of XML are similar " +  myDiff.similar());
    System.out.println("but are they identical? " + myDiff.identical());
}

XML is ordered. They are not the same. Depending on what's reading the file it might behave differently based on the order of the values. — JustinKSU, Mar 11 '14 at 16:10
What is that `Diff` class? I don't see it anywhere in the java core classes. — The Guy with The Hat, Mar 11 '14 at 16:12
@TheGuywithTheHat As the tag says, is XMLUnit: http://xmlunit.sourceforge.net/api/org/custommonkey/xmlunit/Diff.html — Pablo Lozano, Mar 11 '14 at 16:19
@TheGuywithTheHat [Diff](http://xmlunit.sourceforge.net/api/org/custommonkey/xmlunit/Diff.html) is from the XMLUnit API — helderdarocha, Mar 11 '14 at 16:27
I don't think your sample xml is illustrating the issue encountered in your actual code. XmlUnit docs: "Two documents are considered to be "similar" if they contain the same elements and attributes regardless of order." — Brian, Mar 11 '14 at 16:34
@Brian Check my update, there are incoherences between the Javadoc definition and the User Guide. I think the User Guide is the relevant one. — Pablo Lozano, Mar 11 '14 at 22:21

Pablo Lozano · Accepted Answer · 2014-03-11T22:19:25.933

Just guessing, but I think the problem is both tags have the same name. It sounds contradictory, but let me explain it:

<root>
    <field>John</field>
    <field>Smith</field>
</root>

<root>
    <field>Smith</field>
    <field>John</field>
</root>

For me these two pieces of XML are not similar as one says John Smith and the other one says Smith John

<person>
    <name>John</name>
    <surname>Smith</surname>
<person>
<person>
    <name>John</name>
    <surname>Smith</surname>
<person>

These others are similar: not identical but clearly both say John Smith

In other words: as @JustinKSU says, order matters.

UPDATE: From the XMLUnit Java User's Guide:Two pieces of XML are identical if there are no differences between them, similar if there are only recoverable differences between them, and different if there are any unrecoverable differences between them

My second example shows two similar XML pieces because differences are recoverable. The first one isn't because we don't know the correct order: Maybe there is a guy whose name is Smith, so we cannot be sure. Your example is pretty the same case: the parser cannot know if the columns order are important or not. Imagine that your XML is used to select how to order a SQL query:

SELECT * FROM table order by name, id is clearly not the same that SELECT * FROM table order by id, name

I don't why then order matters when I want to only check if they are similar. As your example I don't care what order of fields node is if they are childs nodes of root node. If order would matter for me I would check if they identical ;// . — mardok, Mar 11 '14 at 19:28

Why those two xml string aren't similar

1 Answers1