[RkWard-devel] Length and NAs

Stefan Roediger stefan_roediger at gmx.de
Tue Mar 27 09:05:39 UTC 2007


Am Dienstag, 27. März 2007 01:10 schrieb Thomas Friedrichsmeier:
> On Monday 26 March 2007 23:10, Stefan Roediger wrote:
> > Another suggestion. What about "rk.descriptive.variables"? Might be nice
> > to have in other places too.
>
> Yes, might be nice to have in other places, as well. However, I'd rather
> make it an embeddable plugin, then.
>
> One reason is technical: We need to assign to several different fields of a
> results object (in general a data.frame), and that's not easy to do in a
> single non-cryptic R command.
>
> The second reason is that I'd like to avoid creating convenience function
> that actually "calculate" something. The "core" of the code generated from
> plugins should not have too many rkward specifics, so you can also use it
> in plain R sessions. Using custom convenience functions for providing a
> nice printout of the results after the "real thing" is already done (like
> rk.header(), rk.results()) is something different from actually using
> custom functions to generate the results. (Note that we have a number of
> usages of
> rk.get.description(), that I'd like to move further away from the actual
> calculation, if I only had a good idea, on how to do this, yet).
>
> With an embedded plugin we can solve this easily, as it can simply generate
> plain R statements, that we still won't have to fix up in all plugins that
> use this.
>
> Anyway, something for after 0.4.7.

I see.

>
> For now, I'll assume it's ok with you, if I turn this into a single GUI
> option?

Yes

> Or do you have objections to this change? 

Not really. Basically I myself like to have the options but of course it's 
more convenient for a user to have just a single option.

>
> Regards
> Thomas

Stefan




More information about the Rkward-devel mailing list