El Diumenge, 8 d'abril de 2012, a les 17:13:54, Pau Garcia i Quiles va escriure: > On Sun, Apr 8, 2012 at 4:03 PM, Anne-Marie Mahfouf < > > annemarie.mahf...@free.fr> wrote: > > ** > > > > > > Indeed, Nice idea, I think this is the right focus to (auto)test the > > functionality/features of the app. I've searched some info about this > > topic > > and found this: > > > > http://ldtp.freedesktop.org/wiki/Home > > > > It has full support for KDE/Qt (>4.x) apps and the scripts (for > > autotesting) can be written with Python. > > > > My 0.5 cents :) > > > > Cheers, > > Percy > > > > Yes this is maybe the best free tool to do the job. DO you or anybody have > > used it already? > > Does that tool support QML? Is there an active team behind it? > > Writing UI tests ("functional tests") is a hell of a lot of work and > choosing the wrong tool means in 2 years we may need to maintain the tool > ourselves or rewrite all the tests for another tool. > > I can tell you TestComplete's support for Qt is pretty limited. I have not > tested LDTP because we needed support for Windows and Linux for our Qt > projects. > > Squish is the best tool we evaluated at work for Qt, it does support Qt > Quick and there is a company maintaining it (Froglogic, founded by KDE > developers and employing many KDE developers). A few more arguments > pro-Squish: it's cross-platform (which means we can run the same tests on > Linux, Windows, Mac and any other platform we support) and the > client-server architecture is very useful when testing client-server > applications, actual environments and/or using virtualization to run the UI > tests after each daily build.
Did you guys ever try Testability? I've been using lately and works pretty well and has the added value of being Free Software. Albert > > Talking about virtualization, what we do at work is we have daily builds > for master and stabilization branches and we run tests in virtual machines. > We are currently testing on 12 platforms. We have several "testing > profiles" (test suites) so that we can quickly say the equivalent of "this > version of kdelibs is broken, do not bother performing any further testing: > just flag this build as broken". Everything is automated and launched from > the continuous integration tool as the build finishes. > > I'm a bit out of the testing stuff at work, and very very busy, but if we > are serious about it, I can still give some advice and ask some of the > verification and validation people if they are interested in joining. > > PS: Let's continue the discussion in kde-testing@