[R] some EPS rotated in journal preview

[Marc Schwartz (via MN)]

On Wed, 2006-01-18 at 17:09 +0100, Knut Krueger wrote:
> 
> Marc Schwartz schrieb:
> 
> >On Wed, 2006-01-18 at 13:04 +0100, Knut Krueger wrote:
> >  
> >
> >One of the first things to do is to use 'onefile = FALSE', 'horizontal =
> >FALSE' and paper = "special"'.
> >  
> >
> I am afraid the problem is on the journals side, because the "wrong" 
> postscript line (with "letter" )is working
> and I changed the postscript options between both examples no change in 
> the behaviour of the pdf creator of the journal.
> 
> >This is in the Details section of ?postscript, which provides guidance
> >on the creation of EPS files for inclusion in publications.
> >
> >I would try that to see if that provides a more consistent formatting of
> >the plots.
> >  
> >
> 
> >You don't indicate what you are using to create the manuscript itself
> >(ie. Word, LaTeX,  or ?) to help us in considering other possibilities
> >(such as auto rotation).
> >  
> >
> The journal collects the figures once by once after the manuscript in 
> the original file format.
> to solve the problem I could change it to tiff and submit it, but the 
> tiff files are look ing not as good as eps files.
> 
> >If the above does not help, please provide a reproducible example of the
> >plot code and what you are using for the manuscript.
> >
> for your information:
> the same problem occurs if I plot the data to the graphic device and use 
> the right button -> create postscript file
> One is not rotated the others are rotated.
> 
> Hope the code helps without data:

<SNIP>

Unfortunately, it may have precluded my being able to replicate exactly
what you are seeing, despite being intimately familiar with one of the
functions (barplot2  ;-) that you are using.

I did create some data that would generally fit what you are doing in
both cases, however, probably because I am on Linux and you are on
Windows, where both device and GS differences may be problematic, I
could not get the second to be rotated relative to the page.

If you want, e-mail me (offlist) the rotated EPS file that is created in
the second case and I can use that to try to replicate the problem as
well as review the EPS code to see if something sticks out.

Under Linuxen, one can set:

  GS_OPTIONS="-dAutoRotatePages=/None"
  export GS_OPTIONS

which disables autorotation on the system processing the EPS graphics. I
would envision that there is a similar situation on Windows, using
either an environment variable or via the command line, depending upon
what your publisher is using for tools. A review of the Windows
documentation for GS would be helpful here.

Another possibility is that the rotation is being caused by some
confounding in the text in your second plot. GS will base the rotation
properties on the dominant text in the file. Given your plots, it may be
possible that some of the text label components in the second plot are
causing GS to perform the rotation. If this is the case, disabling the
autorotation as above should help.

The use of 'paper = "special"' is very helpful in general, as it enables
tight bounding boxes around the EPS graphic for inclusion in other
documents. I would highly recommend using that as your default.

One other quick comment, which is relative to the readability of your
code.  Strategically placed spaces (" ") and line breaks would help
tremendously. Especially since most e-mail clients will create line
breaks at 72 chars (or similar), which can make the flow of the code
difficult to review.

HTH,

Marc Schwartz