[Rd] regarding saving R graphis images directly to directory

[Martin Maechler]

>>>>> "Vadim" == Vadim Ogranovich <vograno at evafunds.com>
>>>>>     on Sun, 20 Jun 2004 15:22:41 -0700 writes:

    Vadim> If you just need to ignore the errors use try or tryCatch around the
    Vadim> plotting functions.

    Vadim> Re: jpeg and friends. R has a notion of current
    Vadim> device where it sends all its graphics. If none is
    Vadim> open R opens the default device for you, which
    Vadim> happens to be X11 on your system. To use a different
    Vadim> device just open it, say jpeg, and all graphics will
    Vadim> go there until you close it or open yet another
    Vadim> device. For the list of available devices see
    Vadim> ?Devices

    Vadim> It might be useful to have a null device which just
    Vadim> silently ignores all graphics (aka /dev/null on
    Vadim> UNIX), but I don't know if R has anything like this.


    Vadim> P.S. This sort of questions looks more appropriate for r-help. Just
    Vadim> personal sensing, I am no master of polices.

but you are very right, Vadim.  
Saurin's question would have been appropriate only for R-help.

OTOH, your "P.S." above --- being a proposal for enhancing R ---
does well fit into R-devel.

I agree that it would be nice to have a
  nullDev() or dev.null()
graphics device which would efficiently discard all "plotting to
devices" graphics.  
Note however that it should *not* discard the building of GROBs
(graphical objects) [grid package], i.e. it would construct all
these also for all lattice (or nlme) graphics.  It would just
trash discard when grob's are being 'printed' (i.e. plotted).

A -- quite inefficient -- but easy way on Unix-alikes
{i.e. everwhere but Windows}, 
would be to call, e.g.,   postscript(file = "/dev/null")
I assume there's something equivalent on modern Windows (?)

Martin