[Libreoffice-qa] Regression test (Moztrap) test case localization temporary solution.

[Petr Mladek]

Sophie Gautier píše v Čt 30. 08. 2012 v 21:48 +0200:
> Hi Petr, Yi Fan, all,
> On 23/08/2012 11:59, Petr Mladek wrote:
> > Yi Fan Jiang píše v St 22. 08. 2012 v 21:08 -0600:
> >> we will need to have a temporary workaround to *adapt* translation
> >> version of test cases existed in Moztrap.
> >>
> >> Based on the capacity of current Moztrap, we currently have 3
> >> prototypes of how they can be presented. I populated all these into
> >> three actual test cases:
> >>
> >> http://vm12.documentfoundation.org/runtests/run/29/env/633/
> >
> > I prefer the first version. It best follows the current Moztrap design.
> > The step and expected result are close to each other and better
> > visualized. Also it should be easier to convert this to the real
> > localized framework once we have it.
> >
> > The second solution is more compact but the step and the result is
> > connected by the strange "vpX" identifiers. In addition, it would be
> > more complicated to keep all localizations in sync.
> >
> > The third variant looks to confusing to me :-)
> 
> So the feedback from our team is the 2nd presentation that is preferred 
> over the others, it seems to be more readable for all of them.
> May be we can go for the first presentation until we get more languages 
> but do not forget about another solution where the steps and the 
> expected results are grouped for the same language.

Sophie, thanks a lot for getting the feedback.

When I think about it, also the 2nd approach can be parsed by a script.
It is more complicated, probably more error prone but possible. Thus it
can be moved into to really localized framework almost automatically as
well.

So, the second approach is probably the best temporary solution.


Best Regards,
Petr