[RFC] KUnitTest library

[Frans Englich]

On Thursday 07 April 2005 11:56, David Faure wrote:
> On Thursday 07 April 2005 13:41, Zack Rusin wrote:
> > On Thursday 07 April 2005 05:47, Jeroen Wijnhout wrote:
> > > For me the advantage of KUnitTest is that is integrates well with the
> > > KDE build system, since I'm no automake hero, that's a real plus. I
> > > couldn't get the QtRunner of CppUnit to work (even when I got it to
> > > compile it crashed, while it didn't crash when it was compiled using
> > > the qmake .pro file). So KUnitTest is probably very weak compared to
> > > other tools, but he, it works ;-)
> >
> > I don't think it's weak compared to other tools, in fact I think it
> > stacks rather well against them. Personally I was just never too happy
> > with it. It doesn't make it bad by any means, it just makes it one of
> > the billion things that I wrote and was never fully happy with :)
> > I'm sure you made it substantially better and if you think you'll be
> > able to maintain it, I'd say go for it (put it in kdesdk, or maybe even
> > kdelibs). It would definitely make a lot of sense to have a unit
> > testing library in KDE at some point.
>
> Yes I would really like to have some unit test framework in kdelibs so that
> the unit tests there can all use the same framework.
>
> I haven't tested KUnitTest, but if the lib can be separated from the GUI,
> then let's have the lib in kdelibs and the gui in kdesdk. Would be nice to
> have a non-gui mode for "make check" though.

Yes, I have also an interest in unified unit testing, although I can't 
contribute. I often find myself copying around testing templates which are 
close to not being worth it.

It would be great if it was simple. Nothing over-engineered, but can quikly 
and easy be rigged up, run, and understood. The possibility for flexibility 
and power is nice, but a common scenario such as testing the return value of 
function shouldn't be a lot of work.


Neither no experience of KUT,

		Frans