ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephane Bailliez <>
Subject RE: test cases failing
Date Wed, 26 Dec 2001 08:16:56 GMT
> -----Original Message-----
> From: Erik Hatcher []

> But then why the failure I reported?  Keep in mind that I had not
> specifically enabled forking of tests, so could that have 
> also caused the
> error reported for this one?   It was an IllegalAccessError 
> though.  And it
> passed after I made the change.

I believe it is a classloader issue. I can't give you the details, but I'm
sure Peter is able to.

If you do not enable forking chance is that you will run the tests against
the bootstrap classes rather than the freshly built classes so most of the
time if you don't clean up, they are out of date and result in either old
implementation or invalid interfaces. (it's weird to fix bugs, write tests
and still have the tests failing because you are still testing the old
implementation :)

I made the mistake several times and probably will do it again and again
when I don't pay attention at what I'm doing like forgetting to do a build
bootstrap before running the tests.

Due to this for single test, I'm relying more on the IDEA/JUnit integration
rather than the Ant one(Ant is more appropriate for batch), when I'm using
Ant, I have my with junit.fork=true and my <path
id="classpath"> set to collect the jar in ${lib.dir} so that the forking
works well.

About the batch tests, GUMP does it well. :)

I'm curious to know how other committers are doing though...

> > It is not inappropriate but it is not necessary. :)
> So explain the failure to me then!  :))

I can reverse the question and ask you why GUMP runs the tests correctly for
days, including the SQLExecTest. :-)


To unsubscribe, e-mail:   <>
For additional commands, e-mail: <>

View raw message