[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [geclipse-dev] JUnit Tests
|
Hi,
some q's were already answered but some reain:
> Ariel, you said most of the tests currently fail. Well, at the very
> first page it is stated, that more than 75% of the tests were
> successful, so for me it seems that most of the tests are successful, or
> am I misunderstanding something?
no, i said: when you look at the entry page of the report it looks mostly
red (which means that there is _at_least_ one failed test per package. But
of course each package has several (test) classes and each class has
several tests (methods), so overall the amount of "passed" tests is around
70%
> Furthermore the first page tells me that we currently have 272 tests.
> Does that include really ALL tests (SVN, CVS, Core, PDE, local, Grid,
> ...)?
it includes all tests in all test plugins currently available in SVN+CVS as
they are checked out of the repos. _ALL_ classes found with suffix
_PDETest or _Test are run.
On Wednesday 14 May 2008, Tao, Jie wrote:
> That is true with the Grid tests. The needed stuffs for the Grid stubs
> are curretly not available for the nightly build system. In addition,
> there is still unsolved dependency problem, for example: the Grid
> project can not be created due to: java.lang.NoClassDefFoundError:
> eu/geclipse/ui/wizards/GridProjectProperties
well... but what does that mean, probably that the corresponding test would
also break if i run it from my desktop, right?
Did you look at the exception, why is that happening?
> As for the number of tests: first, not all test plugins are included in
> the automatic system.
?? they are, Markus included them all in the 3 different test-features.
If we missed one please report it to me!
> Second, the statistics of the nightly build system
> is not correct. Just have a look of the second and third columns in the
> test results: in most cases the number of tests is 0 but error number is
> not. So, the number of tests are not correct.
no, if you look at the exception reported in those cases, you will see that
they were thrown in the corresponding "@BeforeClass" method (usually
called "setUpBeforeClass()")
This means the framework did not even get to the point of counting the
tests...
Cheers, Ariel
