Performance of embedded tests and JBPM

[dahm]

Hi,

we've got a lot of embedded tests running with SeamTest as the base class. That is a great feature, however, the tests are very slow.

I found that this is partly because for every seam test class the JBPM process definition we use is redeployed. I.e., the schema is recreated every time and the process definition stored.

Is there any way to avoid this behavior? And do you have
any other optimization hints?

Thanks in advance
  Markus

[tom_goring]

Hi,

I use a custom components.xml when running my integration tests that avoid any jbpm stuff.

That speeding things up for me.

[pmuir]

Try the new SeamTest structure in Seam 2.1 - it's considerably faster to start.

[dahm]

Well, that's not really an option, since we WANT to test the JBPM process and the related issues :-)

I just thought there may be a trick to deploy the process definition just once for all tests.

BTW: We are using Maven/Surefire to start the tests

[dahm]

We are currently bound to Seam 2.0.1. Can I use the new test framework anyway? That is isolated from the rest?

[pmuir]

You can, the underlying structure is the same. Compare the changes in trunk to Seam 2.0.1 to SeamTest.