[R] curiosity question: new graphics vs. old graphics subsystem

Paul Murrell p.murrell at auckland.ac.nz
Tue Jul 4 22:12:48 CEST 2006


ivo welch wrote:
> hi paul:  thank you.  great explanation.
> * what you describe seems like a fairly modest advantage of the
> traditional system.  I have never used the new system, but this part
> of the translation should be simple: if a plot() is called, set the
> current viewport.  emulation done.  (if the user has issued a plot()
> and then screws around with the viewport, R could complain and tell
> the user that she is now on her own.)  I am probably way too naive
> here.  there must be a lot trickier stuff.

There is the package 'gridBase' which provides some support for this
"naive" approach.

> * R regularly makes small changes that break old R code in order to
> improve R.  This is good.  IMHO, emulating the old graphics system
> would not have to be perfect one-to-one.

But the person maintaining the emulation code would have to support more
than just YHO :)

> * when new graphics features are added, there must be duplication of
> effort to get it into both graphics systems.  would it not be easier
> in the long-run to have just one system to maintain?


> * grImport and your slides look nice.  alas, I am searching for
> grImport in your book, and it ain't there.  ;-).  also, pdf should be
> as important as postscript to include these days.

The book preceded work on grImport.

grImport imports PostScript because that is possible without writing a
PostScript interpreter (it uses ghostscript).  Many tools exist to
convert from other vector formats to PostScript (e.g., pdf2ps
[ghostscript], pdftops [xpdf]).

> * speed and memory size for graphics are less of an issue now than it
> used to be.

Right.  Otherwise grid would not exist.  But some datasets and
visualization problems are getting larger at an even greater rate.

> sheesh, this almost sounds ungrateful.  don't mean to be.  thanks to
> the R team for having put this graphics together.  it really is the
> best thing out there.  this discrepancy in graphics systems is really
> nothing more than a minor nuisance.

Suggestions are welcome, as long as they are not accompanied by a
deadline :)  You are of course correct.  It would be much better if
there were only one graphics system.  The only sensible solution would
be an emulation layer, but it needs to become enough of an itch for
someone that they have to scratch it.  At this stage it just seems to be
a bit of a rash;  unsightly, but the cure is worse than the disease.

Dr Paul Murrell
Department of Statistics
The University of Auckland
Private Bag 92019
New Zealand
64 9 3737599 x85392
paul at stat.auckland.ac.nz

More information about the R-help mailing list