Here are my results:

mvn test (dumping output in a logfile):
Running org.apache.uima.cas.test.SerializationReinitTest
Tests run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 9.863 sec
Tests run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 10.057 sec
Tests run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 9.912 sec
Tests run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 9.981 sec

(due to the bug and the different behaviour, I tested mvn install as well)
mvn install (dumping output in a logfile):
Running org.apache.uima.cas.test.SerializationReinitTest
Tests run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 9.88 sec
Tests run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 10.377 sec
Tests run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 9.966 sec

So it is pretty stable around 10 Sec. pretty high compared to the others. If
the claim from apple is true, this would explain the difference. The new
MacBook Pro's are supposed to be 5 times faster ;-) The question however
still remains, why it is slower here than on the other systems. Testing this
in Eclipse brought the same results round about 4 Sec. for the test.

When I tried running all testcases of uimaj-core in Eclipse (right klick on
src/test/java and run as junit test), they not only took forever (felt like
more than 10 min.), there seam to be some real problems. From 327 testcases
tested, 70 of them failed (some ran and failed more than once).
Some examples are these here:
org.apache.uima.cas.impl.CasResetResizeTest
org.apache.uima.cas.test.NewPrimitiveTypesTest
org.apache.uima.cas.test.SerializationNoMDTest
org.apache.uima.cas.impl.XCASDeserializerTest
org.apache.uima.analysis_engine.impl.AnalysisEngineDescription_implTest
org.apache.uima.jcas.test.JCasTest
...

-- Mirko

On 1/17/07, Mirko Jahn <[EMAIL PROTECTED]> wrote:

I will test that on my intel mac as soon as I come home. Maybe there is a
difference between the PowerPC and the Intel architecture versions. As far
as I can remember, the mvn install took about 1:20 and mvn test something
less than a minute for all tests and projects (if that helps).

-- Mirko

On 1/17/07, Adam Lally <[EMAIL PROTECTED]> wrote:
>
> On 1/17/07, Marshall Schor <[EMAIL PROTECTED]> wrote:
> > Somethings is likely amiss - any ideas?
> >
> > Other data points:
> >
> > When running this as part of mvn install - the test on the mac takes
> 57
> > seconds (pretty consistent)
> > When running this from eclipse, as a single test, on the mac, this
> takes
> > 21 seconds( pretty consistent)
> >
> > When running this from eclipse, single test, on the thinkpad, this
> takes
> > 2.0 seconds (pretty consistent)
> >
> > Can someone try this on Linux?
>
>
> Took 2 seconds on the Linux box I use.  No idea why it would take so
> much longer on a Mac.
>
> I wonder if this can be localized to a particular part of the test,
> either with some profiling tool or just old-fashioned
> System.out.printlns.
>
> -Adam
>


Reply via email to