[RkWard-devel] Graphics devices

[Renaud Lacour]

> But I'll try to include acf()
> in the next release. Or do you think it's not desirable?

The problem with including a plugin for ACF plots is that unlike other
plugins, the acf function depends on the optional tseries package.
Moreover, this would require to also include plugins for some related
functions such as pacf or predict.
What interests me especially in Rkward is that it provides me with a
well-featured workspace that R lacks. It allows the user to keep several
displays at one (I mean the R objects explorer, help outputs, graphics
outputs, R terminal and so on). But I do not really use Rkward plugins,
except for package management, because R functions for graphics are often
easy and fast to use.
However I would be quite interested in a simple way to switch
between graphics devices. I'm thinking of a new tab around R terminal tab
which would look like this :

Graphics device
---------------

  [activate] GD1  Filename (if relevant)   GD parameters [Close]
  [activate] GD2  Filename (if relevant)   GD parameters [Close]
  ...

  [Add custom GD]

with the possibility to use a default/automatic GD such as PNG or SVG in
Rkward's output tab.
The aim is to allow both graphics display and file save in exotic formats
such as pictex.
A contextual menu on displayed graphics should be more user friendly
(as in R Windows GUI) butmore restrictive as far as output formats are
concerned, as it acts after the GD.

Well these are only some personnal, maybe not usable, ideas.

> Would you be so kind and provide me some examples (datasets & R
> syntax/images or whatever) of your usage?

I'm doing a school work on the dollar-uro exchange rate time series from
1978 to nowadays. But you can try ts functions on R's own datasets, such
as Nile, sunspots, EuStockMarkets or even white noises (rnorm(1000)),
random walks (e=rnorm(1000); x[1]=e[1]; for(i in 2:1000);
x[i]=x[i-1]+e[i]). Then you can plot these series, draw histograms,
calculate and display ACFs, PACFs and achieve tests to dertermine the
main characteristics of the series, eventually to fit a process
(with eg the arima function) and use the predict function (which displays
a nice graphic) for short term predictions.

BTW, as far as feature requests are concerned, I appreciate the
autocompletion for R objects. Unfortunately, it does not complete file
system paths in functions such as setwd or some graphics devices.

Whatever, thank you for this great pice of software.

Regards,
-- 
Renaud Lacour <renaud.lacour at free.fr>