new hbm2java generated equals() method is completely broken

eepstein · **Posted:** Wed Apr 21, 2004 3:38 pm

I just asked the new hbm2java (extensions 2.1) to generate equals:

Code:

via a non-id property as recommended in the Hibernate pages: see "Seperating object id and business key" on http://hibernate.org/109.html

Code:

   <property name="name" type="java.lang.String" column="name" length="256">
      <meta attribute="field-description">Names are unique and so provide an alternate way to dsrcId for referring to a DataSlice.</meta>
      <meta attribute="use-in-equals">true</meta>
      <meta attribute="use-in-tostring">true</meta>
   </property>

Looking at the generated code I see:

Code:

    public boolean equals(Object other) {
        if ( (other == other ) ) return true;
        if ( !(other instanceof Persistent_DataSlice) ) return false;
        Persistent_DataSlice castOther = (Persistent_DataSlice) other;
        return new EqualsBuilder()
            .append(this.getName(), castOther.getName())
            .isEquals();
    }

And the very first line of code reads:

Code:

if ( (other == other ) ) return true;

Now my question: when will that not be true?! In other words the generated equals() method says this instance is equal to every other object!

Yeah, I know I can write my own Velocity template and guess I'll have to do so, but this looks like a bug to me.

greg_barton · **Posted:** Wed Apr 21, 2004 4:11 pm

I figure it ought to be

Code:

if ( (this == other ) ) return true;

greg_barton · **Posted:** Wed Apr 21, 2004 4:22 pm

Quote:

Now my question: when will that not be true?!

Believe it or not, in an exceptionally rare case, it could be false. Equality testing is not atomic for object references, so if the reference to "other" were changed in another thread at exactly the right time, the test would fail. Granted, it couldn't happen in that particular code snippet (the "other" reference is passed by value to the method, and no threads affecting the reference are spawned before the test in the method, so it's safe) but reference equality tests can fail, even when testing a reference against itself, because reference equality is not thread safe.

Jeez, I'm bored. :)

max · **Posted:** Wed Apr 21, 2004 6:51 pm

please submit a jira for it if it really generates other==other

michael · **Posted:** Wed Apr 21, 2004 7:04 pm

There allready is one at http://opensource.atlassian.com/projects/hibernate/browse/HB-900

immanuel · **Joined:** Tue Jan 13, 2004 4:57 am **Posts:** 83

michael wrote:

There allready is one at http://opensource.atlassian.com/projects/hibernate/browse/HB-900

Will there be another release soon or do we go with the patch?