Testsuite failures for MalformedTestCase testMalformed test

[mwringe]

The org.jboss.test.jbossmx.compliance.objectname.MalformedTestCase testMalformed fails when using the Sun 1.5.0_08 JVM.

A bug report was opened for this (http://jira.jboss.com/jira/browse/JBAS-3549) but was closed (won't fix):
"We can't fix this (short of removing the test).

That is an invalid ObjectName, but Sun's class (unlike JBoss's version)
does not spot the problem."

I was told I couldn't ask questions about the bug in the bug report, and to ask them here. Here are the questions I tried to ask before:

-Could you please explain this issue a little bit more?
-Why does JBoss have separate versions of Sun's classes?
-This test usually passes on the 1.4.2 JVM, is this test only meant to be run on a 1.4.2 JVM?

[ryan.campbell]

Perhaps we should just ignore this if running under jdk5?

[starksm64]

-Could you please explain this issue a little bit more?
A jmx object name string of "domain:=,foo=bar" violates the jmx spec ObjectName syntax of:
[domainName]:property=value[,property=value]*
and ObjectName javadoc which states that

http://java.sun.com/j2se/1.5.0/docs/api/index.html

The key properties are an unordered set of keys and associated values.

Each key is a nonempty string of characters which may not contain any of the characters comma (,), equals (=), colon, asterisk, or question mark. The same key may not occur twice in a given ObjectName.

Each value associated with a key is a string of characters that is either unquoted or quoted.

An unquoted value is a possibly empty string of characters which may not contain any of the characters comma, equals, colon, quote, asterisk, or question mark.

I have submitted a bug report to sun on this.

-Why does JBoss have separate versions of Sun's classes?
We have had our own implementation of JMX since it was an extension package to jdk1.3.

-This test usually passes on the 1.4.2 JVM, is this test only meant to be run on a 1.4.2 JVM?
This test is only meaningful for jdk < 5 to test our implementation.

[mwringe]

Since this test is only meaningful is the JVM version is less than 1.5, shouldn't this test be disabled if a 1.5 jvm is present?

[starksm64]

It would be better categorized as a known failure that can be ignored.