Richard,
This problem has been around for a while. Based on an earlier reply
from Simon on this I find the following has always worked.
quartz.save('tmp.pdf',type='pdf')
Sys.sleep(4)
quartz.save('fig4.pdf',type='pdf')
I recall that plotting twice rather than just the delay helps, but I
don't have notes on that.
As far as machine speed is concerned I don't think it is necessarily due
to slow machines. I had the problem on a slow box (G4 2 procs) and also on
my new box (PowerMac 2.6 GHz, 4 cores, with only idle and system programs
running other than R).
The Quartz output is really nice looking, but many of the journals we
publish in want eps or tiff for final figures. Also, if your co-authors use
Word and not LaTeX then the pdf figures in the manuscript don't look so
great either. The ghostscript conversion usually works, but the quality is
lower than the original pdf. We have found converting to a high resolution
tiff (very large files) is the most reliable way to get figures in journals
without problems.
-John
On 2/13/08 5:53 AM, "Simon Urbanek" <simon.urbanek at r-project.org> wrote:
> Richard,
>
> thanks for the report. The issue is known and the only (known) cure
> lies in R-devel's new Quartz.
> It seems to be triggered by concurrency - e.g. on my machine your
> example produces five equal and correct PDF plots without problems -
> hence it occurs more often on slower machines. You may find several
> post about this bug on this mailing list and some suggestions on how
> to deal with it (adding delays before saving, preventing output and
> probably in your case not starting new devices each time).
>
> Cheers,
> Simon
>
> On Feb 13, 2008, at 7:40 AM, Richard Martin wrote:
>
>> Good Morning,
>>
>> I believe I have found a bug within R.app. It's a problem with
>> quartz.save() which _seems_ to manifest itself when using par(mfcol)
>> to create multi-window graphs (and hence affects topology plots etc).
>>
>> I have created a script which will repeat the bug _most_ of the time.
>>
>> for(i in 1:5) {
>>
>> quartz()
>> par(mfcol=c(2,2))
>> plot(1:10, type="b")
>> plot(1:10, type="b")
>> plot(1:10, type="b")
>> plot(1:10, type="b")
>> quartz.save(paste(i, ".pdf"), type="pdf")
>>
>> }
>>
>> EXPECTED OUTCOME: 5 files, named 1.pdf to 5.pdf, containing 4 graphs.
>> Each pdf should be identical.
>> ACTUAL OUTCOME: The pdf files are most often not equal. Sometimes
>> individual plots are left off the graphs (such that 1.pdf contains
>> only 2 or 3 plots, for example). More seriously, in some instances
>> console text is superimposed in a translated manner over the graphs
>> themselves. This can be seen in the attached pdf.
>>
>> R version information follows:
>>
>> $platform
>> [1] "i386-apple-darwin8.10.1"
>>
>> $arch
>> [1] "i386"
>>
>> $os
>> [1] "darwin8.10.1"
>>
>> $system
>> [1] "i386, darwin8.10.1"
>>
>> $status
>> [1] ""
>>
>> $major
>> [1] "2"
>>
>> $minor
>> [1] "6.2"
>>
>> $year
>> [1] "2008"
>>
>> $month
>> [1] "02"
>>
>> $day
>> [1] "08"
>>
>> $`svn rev`
>> [1] "44383"
>>
>> $language
>> [1] "R"
>>
>> $version.string
>> [1] "R version 2.6.2 (2008-02-08)"
>>
>> R.app version information:
>> R 2.6.2 GUI 1.23 (4932) (4932)
>>
>> Mac Information:
>> Version 10.5.2, build 9C31
>>
>> If you have any idea how to work around this bug, I'd be most
>> greatful. If you need any more information or I can help in any way,
>> please email.
>>
>> Kind Regards,
>>
>> Richard Martin
>> Research Officer,
>> University of Essex,
>> Colchester,
>> Essex,
>> CO4 3SQ
>>
>>
>> --
>> Contendere, Explorare, Invenire, et non Cedere
>> <2 .pdf>_______________________________________________
>> R-SIG-Mac mailing list
>> R-SIG-Mac at stat.math.ethz.ch
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac