[R] Windows, gsview: corrupted bbox of .e?ps graphics

[Michael Friendly]

System: [Windows XP Pro, R 2.8.0 and R 2.81.]

When I do R graphics on Windows and want .eps output for LaTeX, with *tight
bounding boxes* I usually do as follows from Rgui:
 - Save as Postscript ... myfig.ps
 - Open myfig.ps in Gsview, select PS to EPS, Save as ... myfig.eps
This always worked before, and gave me .eps files that did not include much
more white space than I like, didn't require fiddling with par(), nor using
\includegraphics[trim=....]{myfig}.

But now I've encountered a problem that prevents this workflow from
working --- when I now open *any* R-generated-this-way .ps or .eps file,
with GSview (v 4.9), the bounding box of the graphic is corrupted--
i.e., the bounding box I see cuts off the top half of the figure.
See:
http://euclid.psych.yorku.ca/SCS/Private/agree-MS-buggged.eps
for
the example produced below.

This does view OK with gv on linux, however, the usual tool I use on
linux to find new, tight bbox won't work on this file:

%psfixbb -l agree-MS-bugged.eps
psfixbb: Probloem with new %%BoundingBox: -4 -4 616 796

What caused this was an example from vcd, that I wanted in .eps form.  
What I got from
my usual method of Save as Postscript was not at all pleasing--- font 
sizes in the
.ps image were way larger than what appeared on the screen.  So, I added a
postscript() command to the example in ?agreementplot

library(vcd)
data("MSPatients")
## best visualized using a resized device, e.g. using:
#get(getOption("device"))(width = 12)
dev.new()

postscript(file="agree-MS.ps", paper="special", horizontal=FALSE, 
width=6.0, height=3.0)
pushViewport(viewport(layout = grid.layout(ncol = 2)))
pushViewport(viewport(layout.pos.col = 2))
agreementplot(t(MSPatients[,,1]), main = "Winnipeg Patients",
              newpage = FALSE)
popViewport()
pushViewport(viewport(layout.pos.col = 1))
agreementplot(t(MSPatients[,,2]), main = "New Orleans Patients",
              newpage = FALSE)
popViewport(2)
dev.off()

Ever since, when I use gsview on Win to view *any*  .eps file  
generated  via Save as Postscript from the R graphics window,
I get similar behavior. This happens in a fresh R session, as well as 
with R 2.8.1.  It doesn't happen with Sweave-generated
graphics (but I haven't tried the above example in Sweave).

Any ideas on what is wrong and how I can fix this?

-Michael

-- 
Michael Friendly     Email: friendly AT yorku DOT ca 
Professor, Psychology Dept.
York University      Voice: 416 736-5115 x66249 Fax: 416 736-5814
4700 Keele Street    http://www.math.yorku.ca/SCS/friendly.html
Toronto, ONT  M3J 1P3 C
NADA