[RkWard-devel] Length and NAs

[Stefan Roediger]

Am Montag, 26. März 2007 20:33 schrieb Thomas Friedrichsmeier:
> On Monday 26 March 2007 00:08, Stefan Roediger wrote:
> > Am Montag, 26. März 2007 00:42 schrieb Thomas Friedrichsmeier:
> > > On Sunday 25 March 2007 22:02, Stefan Roediger wrote:
> > > > So far I did not put any work in the others but plan so for the
> > > > coming week. However, for the others in "TEST" and "Variances" I can
> > > > say that they work but do not use all of the latest development (e.g.
> > > > local- and NA-issues).
> > >
> > > Are those feature complete?
> >
> > Yes, in my opinion they are.
>
> Ok, I've looked at those as well, and adjusted them for local().
>
> For bartlett and fligner tests, I have modified them to accept a list of
> vectors instead of one vector and a grouping factor. I think it's more
> consistent with what we use elsewhere, and acutally a bit easier to do. In
> the mid term, we'll probably want to add an interface for grouped data a
> least for the most important few tests.

Maybe a good idea.

>
> Other than that, mostly minor stuff. Could you take another look to see,
> whether I've broken anything?

I checked them and get the same (complete) results as for console driven 
tests. For me they are okay. Better handling of error messages needs to be 
addressed in future.

>
> Any idea on where to put / how to label "TESTS", yet, anybody?

Still no idea on that here. For the moment being we could also put them in the 
analysis main menu till related test emerge and a new "home" can be 
established.

>
> On another note, I've finally eradicated all uses of "rk.temp" variable
> names and "cleanup()" functions. If I can find the time, I'll move some
> more plugins out of under_development, and prepare a first preview release,
> tomorrow.
>
> Regards
> Thomas