I think we're moving away from having one big browsertests repository at this point to having the tests in the thing they are testing. I know that is the case for extensions but I'm not sure about core.
I try to target my tests so they'll run either on beta or an install with the "required" plugins installed. I keep pretty good documentation about how you go from a fresh install to the required setup. I wonder if we could give the tests some control over a ragrant instance so they could actually set things up the way you'd like. Nik On Thu, Jan 16, 2014 at 9:02 PM, Chris Steipp <[email protected]> wrote: > Hi, I'm pretty new at Selenium and ruby, but I'm working on getting some > tests to work against our MediaWiki release branches, so we can have > automated regression testing for our releases. > > I've noticed that a lot of the tests in the browsertest repo are written > for certain configurations, typically what's on beta labs. As a result, > these tests don't work on a fresh install of MediaWiki. For example, > several tests that look for an "Edit" link look for "Edit source", i.e. VE > is installed. > > At the moment, I've cloned into another repo and I'm just fixing the tests > to work for a tarball install, but is there a proper way to extend those > tests so that we don't need an entire new repo for VE vs. non-VE tests? I > see several places that load a different page object for ie6/phantomjs, is > that a good design pattern that I should copy? > > Thanks for the help! > > _______________________________________________ > QA mailing list > [email protected] > https://lists.wikimedia.org/mailman/listinfo/qa > >
_______________________________________________ QA mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/qa
