Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [technology-pmc] [eclipse.org-planning-council] MPC requesting a Photon simrel respin



On Tue, Jul 3, 2018 at 9:04 PM, Carsten Reckord <reckord@xxxxxxxx> wrote:
And as I commented on bug 536502, I think it would be good to enable projects to push integration tests to simrel to automatically check that they work properly in the final packages together with all the other bits. I'm happy to help with that in any way I can...

What are the technical blockers for this to happen? EPP/SimRel can reference repositories containing test cases, create JUnit test suite combining multiple tests, use Tycho to run those tests on fully provisioned p2 products... Am I missing something on the technical side? Or is it now just a matter of someone doing this at least for a dummy test in EPP to setup the trend?
I think it's not really the role of projects to push for some tests to guarantee things work in EPP; it's more the other way round: EPP to identify and pull (and even produce) meaningful tests to validate the intergration. And I actually think that the quality of modularity in Eclipse world makes those tests not so relevant and that it would still be more profitable. By not relevant, I'm mostly wondering how likely it is that EPP finds a bug that's based on the conjunction of projects running a project test suite? I didn't verify closely, but I believe that out of the latest EPP/SimRel bugs, there are not many that would have been trivially found automatically by running an existing test (launcher name wouldn't have been, MPC issue with path neither, and moreover are not caused by integration of multiple components; maybe the only things we'd spot would be issues with conflicting versions of guava...).
So while I'm all in favor of more automated tests in EPP, it's also important to realize that it's not really something that would avoid most bugs we've seen lately.

Cheers,

Back to the top