[rkward-devel] Second call for testing: RKWard 0.5.3-pre2

[Thomas Friedrichsmeier]

Hi,

On Tuesday 20 April 2010, mat wrote:
> Thomas Friedrichsmeier a écrit :
> > could you post/attach the difference listings for these tests?
> 
> copy/pasted on attached file.  It is a little bit messy as some text is
> hidden in console.

thanks. Let's see:

> lillie_test      
> shapiro_wilk_test  

Some noise in the last few digits. We'll need to find an elegant way to have 
those properly rounded off in the output. Not easily fixable right now, and not 
a real problem.

> kpss_test 

Apparently the test produces unstable results for very short sequences (10 
values). I changed the test to use longer sequences, only. Please try again 
after an SVN up (or with the next preview, once it is available).
 
> Wald_test 
> RSM_parameter_estimation 
> PCM_parameter_estimation 

Meik, can you comment on those (if you are currently available)?
 
> import_spss           
> import_stata    

These are interesting. Could you toy a bit with the SPSS-import plugin and / 
or read.spss() to see if there is any way to import the sample .sav-file (in 
the tests/-directory)? This file has some variables with non-latin1-characters 
in their names, and that's where it fails, apparently. It works on this 
system, however. Maybe there is a way to support this in the plugins.

> density_plot

A harmless message. I'm not quite sure how to make the test ignore this, yet.

> Is there a possibility:
> -to have output separately from console so everything shown and easier
> to send you?

Prasenjit took care of that in SVN. The next preview release will have that.

> -to run tests separately?

Not quite (although this should be possible to implement, sooner or later). 
You can run the "suites" separately, though:
- Open one of the test scripts in /your/rkward/sources/tests/ (e.g. 
analysis_plugins.R)
- Run->CD to script directory
- Run->Run all

> so this would be the list:
> install.packages(c("R2HTML", "tseries", "nortest", "outliers",
> "exactRankTests", "moments", "car", "hdrcde", "qcc", "xtable", "eRm",
> "ltm"))

Thanks for compiling the list.

> maybe you could incude it in the compiling/testing website and/or in the
> plugintest makefile itself?

make plugintests now displays a notice with the list, in case any libraries 
were missing.

> I tried to used this. But as soon as I use "ctrl", brings me back to
> latest console line... 

Indeed. Fixed, now.

> About which should be the default, as rather "GUI
> user" and so linux geek, I would argue ctrl+c as copy is more intuitive...

Well, so now we have one vote for changing it, and one vote for keeping it, 
and I can see the point in both. I suppose this means it will remain as is. 
However, I guess it should be possible (for 0.5.4) to show a notice when the 
user first tries Ctrl+C in the console. Now that you can *really* change the 
shortcut, that should be good enough for everyone.

Regards
Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/rkward-devel/attachments/20100421/ee19e375/attachment.sig>