Trellis graphics and clipping

Even when displayed in the R Graphics Window, there is
some clipping (the extreme top, bottom, left and right
edges are not there, so that labels, and parts of numerical
scale markers all round, have been clipped off).
[snip]
When I switch to the "postscript" device, with the comamnd
'postscript(filename,...)', the result is even more clipped,
[snip]
I've tried playing with all the obvious (to me) parameters
which I can find, but it has not effect on this.

On 29-Jul-01 Ted Harding wrote:

Even when displayed in the R Graphics Window, there is
some clipping (the extreme top, bottom, left and right
edges are not there, so that labels, and parts of numerical
scale markers all round, have been clipped off).
[snip]
When I switch to the "postscript" device, with the comamnd
'postscript(filename,...)', the result is even more clipped,
[snip]
I've tried playing with all the obvious (to me) parameters
which I can find, but it has not effect on this.

Follow-up.

I've tried the following:

  hgm<-histogram( ~N | B , xlim=x(0,5), layout=c(5,5) , ... )

  print.trellis( hgm , position=c(0.075,0.075,0.925,0.925) )

so as to force down-scaling of the graphic, and now everything
is fully displayed in X11: top and bottom scale annotations,
bottom label and side label, and all the histograms. No clipping
whatever. No problems.

Except that there's an error message:

  Error in unset.gpar(vps[[1]]$gp) : trying to reset with 10

Next, I've done

  postscript(horizontal=FALSE,onefile=FALSE)

and again

  print.trellis( hgm , position=c(0.075,0.075,0.925,0.925) )

(getting the same error message)

and then looked at the resulting PS file. This is again slightly
clipped (on the left, as it happens, so that the left label lacks
the tops of its letters), but everything else is absolutely fine.
OK,

  print.trellis( hgm , position=c(0.1,0.1,0.925,0.925) )

(same error message).

Now, Lo and Behold, when I view the PS file there is no clipping
at all. Good!

So, apart from this error message (which must mean something, but
nevertheless goes with perfect output) all now seems fine.

But it still puzzles me that one has to grope for the correct
"position" parameter settings for each device.

Is there a way to pull all this together?

Hi again folks,

I seem to have got the trellis (lattice + grid)
business
working basically OK, in that I can get my 5x5 array
of
histograms laid out as they should be.

However, there is a behaviour I can't see my way
round.

Even when displayed in the R Graphics Window, there
is
some clipping (the extreme top, bottom, left and
right
edges are not there, so that labels, and parts of
numerical
scale markers all round, have been clipped off).

When I switch to the "postscript" device, with the
comamnd
'postscript(filename,...)', the result is even more
clipped,
in that parts of the histogram plots themselves get
cut off,
including their titles as seen in the X11 display.
On the
other hand (using gv's "position meter"), it seems
that the
internal %%BoundingBox matches the dimensions and
coordinates
of what SHOULD get displayed/printed, allowing
plenty of
"white space" round the bit of the graphic that
_does_ get
displayed, so that the whole thing could have been
fitted in
anyway.

And if I go at it crudely, simply using
'dev.print()' to
bang out a har-copy, it is even more clipped again
...

I've tried playing with all the obvious (to me)
parameters
which I can find, but it has not effect on this.

I've checked the PostScript file itself, and the
parts that
are missing are not mentioned therein, so it's not a
PostScript
"clippath" phenomenon: it's definitely R which is
deciding not
to output them.

What's needed is some way of coordinating the
dimensions of
the various levels of this process, so that the
output to X11
is scaled so that the whole graphic can fit in, and
the output
to postscript is also scaled so that it fits into
whatever
size of graphic is nominated (not that this should
be an
issue for EPS output anyway -- it would be quite
sufficient
to output everything and then put a BoundingBox
round the
lot, and never mind things like papersize and so
forth).

So I'm bewildered. And, if anyone knows a way round
it,
I'd be most obliged to hear!

With thanks,
Ted.

E-Mail: (Ted Harding) <Ted.Harding at nessie.mcc.ac.uk>
Fax-to-email: +44 (0)870 167 1972
Date: 29-Jul-01                                     
 Time: 13:40:47
------------------------------ XFMail
------------------------------

r-help mailing list -- Read
http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-help-request at stat.math.ethz.ch