[R-SIG-Mac] Resent with pdf screen shots: Problem with quartz graphics on macpro

[Ashwin Kapur]

Hi all:

Sorry for the resend but after re-revieiwing the mailing list guidelines, it seems I'm only supposed to attach pdfs as attachments so I'm resending with the screenshots converted to pdfs as opposed to pngs.

I'm having a strange issue with quartz graphics on my mac pro desktop.  I'm really hoping someone has seen this and knows what's going on because I'm at my wits end.

First the details of the machine.

uname -a
ashwin macpro 10.2.0 Darwin Kernel Version 10.2.0: Tue Nov  3 10:35:19 PST 2009; root:xnu-1486.2.11~1/RELEASE_X86_64 x86_64

R version 2.10.1 (2009-12-14)
R 2.10.1 GUI 1.31 Leopard build 64-bit (5537)
The macpro boots to the 64 bit snow leopard kernel.

One fact which might be important is that I have a dual monitor system but the two monitors do not have identical resolution.  The primary monitor is actually a 42 inch 1080P television and the resolution listed by the OS is 1080P (Television).  The secondary monitor is an old Silicon Graphics monitor driven by a SGI Multilink adapter at 1600x1024 which sits on the right and closer to me which I use primarily for reading documentation/papers etc which I prefer to monitors available today because reading it is like reading from paper.

I run R on two macs, my macbook and my macpro.  The problem only happens on the macpro and only with quartz graphics.  If I do an x11() and then plot to the x11 window, the problem does not occur. It may be easiest to show the problem with screenshots so I'll post them.  Hopefully this isn't a breach of mailing list etiquette.

To reproduce the problem all l I need to do in a freshly started R GUI console is the following.

a <- rnorm(100)
plot(a)

I can also do this via the command line in a terminal by doing.
a <-  rnorm(100)
quartz()
plot(a)

I get the below.  The plot is absolutely huge and is clipped below and on the right.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: SS1.pdf
Type: application/pdf
Size: 10783 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mac/attachments/20100314/84f7702e/attachment-0004.pdf>
-------------- next part --------------



If I resize the quarz window by dragging the bottom right window, the clipping goes away and subsequent plots are not clipped, but notice how the margins and fonts used are huge.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: SS2.pdf
Type: application/pdf
Size: 18115 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mac/attachments/20100314/84f7702e/attachment-0005.pdf>
-------------- next part --------------



I cannot reproduce the problem on my macbook.  I also don't get the problem if I use x11() to start up x11 graphics, I get graphics very similar to what I get if I plot on one of the machines on my ubuntu based cluster.


Below for reference is the output plot(a) produces when run using the x11() device.  This is very similar to the output produced on the quartz device on my laptop.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: SS3.pdf
Type: application/pdf
Size: 16832 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mac/attachments/20100314/84f7702e/attachment-0006.pdf>
-------------- next part --------------



Another interesting thing to note.  When I do demo(graphics) from the R GUI Console and it gets to 

coplot(lat ~ long | depth, data = quakes, pch = 21, bg = "green3")

It produces the error "Error in plot.new() : figure margins too large"  and the output:

-------------- next part --------------
A non-text attachment was scrubbed...
Name: SS4.pdf
Type: application/pdf
Size: 29110 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mac/attachments/20100314/84f7702e/attachment-0007.pdf>
-------------- next part --------------



I've googled/browsed the list etc.  One additional issue I notice is that 

par()$family is NULL, which was something other had been asked to check but setting par(family="mono") doesn't make any difference.

Any suggestions, help, pointers etc would be greatly appreciated.