[R] Encapsulated postscript and the family argument

[Prof Brian Ripley]

On Tue, 26 Aug 2003, Patrick Connolly wrote:

> On Mon, 25-Aug-2003 at 08:03AM +0100, Prof Brian Ripley wrote:
> 
> |> On Mon, 25 Aug 2003, Patrick Connolly wrote:
> |> 
> |> > > version
> [...]
> 
> |> > However, what wasn't obvious to me was that it is necessary to specify
> |> > what family to use.  If no family is specified, the default family
> |> > does appear to be used, BUT, the resulting file is no different from a
> |> > 'regular postscript' file.  The value in ps.options does not seem to
> |> > be used in the same way.
> |> 
> |> The family used is nothing to do with EPS.  The code is always
> |> EPS-conformant (but may not be a single page), but the *header* is only
> |> sometimes, the times being documented.
> |> 
> |> > Is this intentional behaviour?
> |> 
> |> Is what, exactly?  
> 
> A.  What I thought was going on with family seemily being used
> differently when paper was 'special'.
> 
> Now with some gentle prodding, I see that I was confusing two
> different plots I was working on.  Some swapping back into memory
> hadn't finished on Monday morning when I made my observation.  All
> rather embarrassing.
> 
> 
> |> A long-timer such as yourself really, really should know not to
> |> send in vague statements not backed up by the code used to leap to
> |> these conclusions!
> 
> Yesterday, I knew considerably less about the difference between EPS
> and regular PostScript (and most of that was misconception), so I was
> unaware how simple the distinction was.  Constrained by that
> ignorance, I couldn't think of a way of showing more clearly what I
> was on about.
> 
> Thanks Brian for your patience in helping me sort that out.
> 
> I have a small question about that difference:
> Am I correct now in thinking that apart from the first line of a
> single page graphic file (with current versions) reading
> 
> 	%!PS-Adobe-3.0 EPSF-3.0
> 
> instead of
> 
> 	%!PS-Adobe-3.0,
> 
> the only substantial differences between an EPS and a PS file are the
> positioning of the origin of the bounding box at 0, 0 and the removal
> of page orientation information?

Not the bounding box: EPS files can have a non-zero origin (although it is 
not very useful).  The header and the lack of the orientation comment are
the key: the latter is somewhat ambiguously defined, and version 6.0 
ghostscript started rotating figures to have height > width if it were 
included.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595