RE: Unit Test Failures

2012-10-31 Thread Edison Su
o: cloudstack-dev@incubator.apache.org > Subject: Re: Unit Test Failures > > Hugo, > > You are correct -- I missed a very important word there. An issue with using > an in-memory database for integration tests are the differences between > SQL dialects (i.e. a valid query for HSQLDB may no

Re: Unit Test Failures

2012-10-31 Thread John Burwell
Hugo, You are correct -- I missed a very important word there. An issue with using an in-memory database for integration tests are the differences between SQL dialects (i.e. a valid query for HSQLDB may not be valid on MySQL -- degrading the value of the test). Also, have considered using T

Re: Unit Test Failures

2012-10-30 Thread Hugo Trippaers
I think/hope that you mean 'should not require' :-) I think we can do something with the integration test framework in maven. I haven't looked at it in detail yet. Ideally even that should by sort of self contained by using an in memory database during the integration test fog example. For no

Re: Unit Test Failures

2012-10-30 Thread John Burwell
Hugo, Generally speaking, unit tests should require any outside resources -- including a database. Is there a way we can separate the tests that require a database into an integration test suite? Such an approach would preserve the important testing being performed by these tests to continue,

RE: Unit Test Failures

2012-10-30 Thread Hugo Trippaers
Hey John, The unit tests disabled in the maven build for a long time, because a lot of the unit tests do indeed require an active database. Today I disabled most of those database tests and enabled the test procedures in maven as I don't think every developer has an active database, but still m

RE: unit-test failures due to changes in ComponentLocator

2012-08-07 Thread Alex Huang
Thanks. Hopefully it's fixed now. --Alex > -Original Message- > From: Ewan Mellor [mailto:ewan.mel...@eu.citrix.com] > Sent: Tuesday, August 07, 2012 5:28 PM > To: cloudstack-dev@incubator.apache.org > Subject: unit-test failures due to changes in ComponentLocator > > Alex, > > The fol

RE: Unit test failures

2012-08-06 Thread Ewan Mellor
.testGetRandomIpFromCidr(NetUtilsTest.java:53) Thanks, Ewan. > -Original Message- > From: Mice Xia [mailto:mice_...@tcloudcomputing.com] > Sent: Monday, August 06, 2012 1:17 AM > To: cloudstack-dev@incubator.apache.org > Subject: RE: Unit test failures > >

RE: Unit test failures

2012-08-06 Thread Mice Xia
By adding some Daos/Managers to component locators, I can pass SecurityGroupManagerImpl2Test and CheckPointManagerTest But ResourceManagerTest is tough. It sets almost everything up without using Mock objects, which means a lot of managers/Daos and their cascading dependency need to be added in