Hi Steve, this all sounds very reasonable and I'm really happy that we finally get started with testing:) One small comment regarding the platform names: I think we should really call it x86_64 (and I suppose that is what you are speaking about) to avoid any confusion with Itanium which is commonly abbreviated as IA64. Volker On Mon, Nov 26, 2012 at 8:44 AM, Steve Poole <spoole@linux.vnet.ibm.com> wrote:
hi all -
I'm in the process of getting jtreg tests running against the port builds. Running the tests seems straightforward.
I want to brain dump my thoughts on the testing approach for using jtreg and work out how wide a range of testing we need to do.
Ideally I would want to have tests run on:
1 - The powerppc port of Linux PPC 2 - The powerppc port of AIX 3 - The powerppc repo build on Linux Ia64 4 - The powerppc repo build on Windows Ia64 5 - The powerppc repo build on Solaris 6 - The powerppc repo build on Mac OSX 7 - The powerppc port of Linux PPC + IBM J9 JVM 8 - The powerppc port of AIX + IBM J9 JVM
And be able to compare the results with the mainstream JDK7 and JDK8 versions
That could be a lot of testing. (especially since it takes 24hr+ to run the tests on LinuxPPC right now!) I think its ok at this stage to drop the Windows , Solaris and Mac builds and revisit them later. That gives us 5 sets of builds to test. That seems fairly straightforward.
I'm getting some education today on the intricacies of Jtreg curtesy of some knowledgeable Oracle folk - hopefully I can tackle the question of what tests to run and how to compare results over time.
The final thought I had was that we'll need a plan get the ppc repos updated to be near current with the mainstream repos and keep them fairly close so that comparing tests across repos is fairly reasonable. We can do that later but we do need to consider it.
Cheers
Steve