Has anyone ever thought of maybe implementing some other testing framework - maybe testNG? I have managed to run a JUnit4 based test (with lots of changes in the source code and heaps of hacks) - and I can see lots of problems on the way... JUnit4 works very different compared to JUnit3 and from what I can see, it might be very hard to make JUnit4 and GWT work together nicely. I'm sure there would be a way but I don't think it would be a nice and easy one...
The other solution might be to mimic JUnit4 style in GWT TestCases.... On Apr 19, 7:35 pm, Marko Vuksanovic <[email protected]> wrote: > And I've deleted patch set which contained deleted jars. > > On Mon, Apr 19, 2010 at 5:23 PM, Marko Vuksanovic <[email protected] > > > > > > > wrote: > > I've added a patch set 3 which contains patch for tools folder with only > > junit4 jars added - junit3 jars still there. > > > On Mon, Apr 19, 2010 at 4:38 PM, Freeland Abbott <[email protected]>wrote: > > >> On Sat, Apr 17, 2010 at 8:24 AM, Marko Vuksanovic < > >> [email protected]> wrote: > > >>> Here's the patch which replaces the jars (patch set 1) in the tools > > >> Don't delete the old jars, just add the new ones. That way, regardless of > >> which version of source you check out, the tools directory will have what > >> it > >> needs. > > >>> folder and another (patch set 2) which updates build.xml and classpath > >>> files. > > >> This one looks okay, but clearly can't be submitted without the first one. > > >> The junit dependency is so widespread it might make sense to define a > >> macro for it in common.ant.xml. It won't save you anything this time, > >> since > >> you have to update all the cites to that macro, but it might be convenient > >> for next time. > > >>>http://gwt-code-reviews.appspot.com/334803/show > > >>> I'm not sure if I should have opened two different issues, with > >>> different base URLs, for these two patches.... > > >> Typically, the multiple patch sets are opened for revisions of one atomic > >> change. So I'd've done this with two changes (tools and trunk), but you > >> might also have done it with one at the googlecode.com/svn level), but > >> two patch sets with different svn roots and order dependencies between them > >> is odd. > > >> On Apr 16, 10:49 pm, Marko Vuksanovic <[email protected]> > >>> wrote: > >>> > Ok, I'll just save you some time... :) but will first check if > >>> everything is > >>> > ok once the dependencies are replaced... > > >>> > On Fri, Apr 16, 2010 at 10:46 PM, Freeland Abbott <[email protected] > >>> >wrote: > > >>> > > I think that's correct, but we historically have been pretty lazy > >>> about > >>> > > updating our dependencies... > > >>> > > On Fri, Apr 16, 2010 at 4:39 PM, Marko Vuksanovic < > >>> > > [email protected]> wrote: > > >>> > >> One more question - why aren't JUnit 4 libraries used? I think > >>> junit4 > >>> > >> supports junit3 test style? Am I right? > > >>> > >> On Fri, Apr 16, 2010 at 10:12 PM, Marko Vuksanovic < > >>> > >> [email protected]> wrote: > > >>> > >>> I like challenges :):) > > >>> > >>> On Fri, Apr 16, 2010 at 10:09 PM, John Tamplin <[email protected]> > >>> wrote: > > >>> > >>>> On Fri, Apr 16, 2010 at 3:45 PM, Marko Vuksanovic < > >>> > >>>> [email protected]> wrote: > > >>> > >>>>> I just noticed that there is HasAnnotations interface... I might > >>> try > >>> > >>>>> putting that to use... > > >>> > >>>>> On Apr 16, 9:38 pm, Marko Vuksanovic <[email protected]> > >>> > >>>>> wrote: > >>> > >>>>> > Well I've been looking at the existing code and I have seen > >>> that > >>> > >>>>> > JClassType implements getMethods(). Then I would be able to > >>> check if > >>> > >>>>> a > >>> > >>>>> > method has Annotation applied - using isAnnotationPresent() > >>> method. > > >>> > >>>>> > I am not sure how to check if a method has an annotation > >>> applied and > >>> > >>>>> > not to use the refelection package. The problem is that that > >>> package > >>> > >>>>> > cannot be used in translatable code. Obviously I'm missing > >>> something > >>> > >>>>> > here... > > >>> > >>>> You probably want to look at GWTRunnerGenerator, which generates > >>> the > >>> > >>>> implementation of GWTRunner to actually run it (and has access to > >>> > >>>> TypeOracle), or JUnitShell, which actually drives running the > >>> tests and can > >>> > >>>> use reflection to look at the annotations (though be careful about > >>> different > >>> > >>>> classloaders). > > >>> > >>>> That said, I suspect supporting JUnit4 is likely to be a lot of > >>> work. > > >>> > >>>> -- > >>> > >>>> John A. Tamplin > >>> > >>>> Software Engineer (GWT), Google > > >>> > >>>> -- > >>> > >>>>http://groups.google.com/group/Google-Web-Toolkit-Contributors > > >>> > >> -- > >>> > >>http://groups.google.com/group/Google-Web-Toolkit-Contributors > > >>> > > -- > >>> > >http://groups.google.com/group/Google-Web-Toolkit-Contributors > > >>> > --http://groups.google.com/group/Google-Web-Toolkit-Contributors > > >>> -- > >>>http://groups.google.com/group/Google-Web-Toolkit-Contributors > > >> -- > >>http://groups.google.com/group/Google-Web-Toolkit-Contributors > > --http://groups.google.com/group/Google-Web-Toolkit-Contributors -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
