Ming Zhang wrote:
> Hi Sahoo,
>
> Ideally, the failure in war deployment should fail the current test
> only as you mentioned. But knowing all tests requiring war deployment
> will fail at the same place any way,
I don't understand how you conclude that one test failing to deploy
implies all tests will fail to deploy. Can you elaborate? Each
application is different, so the deployment backend can reject one
application yet accept another. If all tests fail to deploy, then so be
it, at least we will see that in one QL run as opposed to seeing it
incrementally.
> we will not see a big difference.
We will.
> The current "failonerror" deployment will cause QL not to run the
> tests. This is not an ideal implementation and I'll put this in my QL
> improvement list. It'll also be good if we can add some deployment
> tests in QL to isolate the deployment failure.
Even if we add some tests for deployment area, that can't guarantee that
every app will be deployed after that test runs successfully. We need to
treat each deployment separately and skip only that test. IMO, this is a
must have.
>
> Thanks for providing a function list for a "reasonably good test
> framework". I am working on improving QL and running single test of QL
> will be in my top list since a lot of people mentioned that. I believe
> TestNG has most of the function in the list and I will work on to
> fully utilize the TestNG features. If you have any specific suggestion
> to improve the framework, please let me know.
>
We don't have to use TesntNG to do all these, but that's your domain.
Good luck.
Thanks,
Sahoo