Hi there all, @Sebastian: it might look a bit extreme, but I believe that at the moment it might be easier to get at least 5 of the community members investing 1 day per week to run automated tests on PRs, and hopefully merging them, than get 1 member to work 100% having a proper CI in place.
Please don't get me wrong: CI is important and we need that! I just don't see how the community can collaborate to get it done within a couple of weeks. For example, how could I, with 5-6 hours per week, help on that? I know that with such time I can help reviewing starting automated tests on a couple of PRs. @Jeff: yes, perhaps having to test a PR before creating one is too much, because not everybody has a test environment to run automated tests. However, the "nobody likes testing" should not be the way we start with this idea. What if we try the following: * If the person creating a PR has tested his/her changes either manually/automated with simulator/hardware and shows some evidences, then having a review + integration tests from a comm member - whilst CI is not done, should be enough to get a LGTM and the PR should be tested. But a LGTM on code review only should not suffice. For UI changes that have been tested by the author with screenshots on the PR, a code review with 1 LGTM should be enough to merge it. What do you guys think? Cheers, Wilder Sent from my iPhone > On 19 Feb 2016, at 19:43, Ramanath Katru <ramanath.ka...@citrix.com> wrote: > > Can we get the CI proposed by Bharat up onto the ASF servers? While may not > be complete or even if its running with issues, why not start it over there > and have everyone fix it to get it up and running? > > Ram Katru > > > -----Original Message----- > From: sebgoa [mailto:run...@gmail.com] > Sent: Thursday, February 18, 2016 2:07 PM > To: dev@cloudstack.apache.org > Subject: Re: [Proposal] Concerning open PRs > > >> On Feb 18, 2016, at 8:38 AM, Wilder Rodrigues >> <wrodrig...@schubergphilis.com> wrote: >> >> Hi all, >> >> >> It’s been a long time, but the Wolverine is not dead yet. ;) >> >> Currently we have 175 opened PRs, which we all agree to be a lot, given the >> fact that few people, if any, are testing/merging them. I have been a bit >> off the radar, but from next week I will start helping to get some of those >> PRs tested and, hopefully, merged. >> >> In order to get the community working as an unit, I would like to propose >> the following: >> >> * One should only create a PR after testing an existing PR. >> - By testing I mean… testing. Not just looking into it and saying “LGTM”. >> Manual tests should also count, with screenshots attached to the PR. >> >> That will make those with test environment pitch in and help, and in >> addition might also decrease the frenzy for creating PRs which occasionally >> won’t be tested within a month time - or longer. >> >> For others not creating PRs that often, like me, we should help testing at >> least 1 PR per week. >> >> Being a bit more blunt now, if a PR is created but the person does not >> contribute with testing an existing one, the new PR should be closed. >> >> What do you think? > > That sounds too extreme to my taste. > > Bottom line, folks who submit PR need to stay on top of them and address > comments. > > But then we need to have that CI in place > > -sebastien > >> Cheers, >> Wilder >
