[changeset] improved support for pdf output

[Ben Abbott]

On Apr 8, 2009, at 2:17 PM, Michael D. Godfrey wrote:

> Ben,
>
> Here is what I learned about matlab behavior:
> matlab version 7.5.0 (R2007B)
> I did:
> print -depsc  't2eps.eps'     -> %!PS-Adobe-2.0 EPSF-1.2
>
> print -dpdf  't2pdf.pdf'        -> %PDF-1.4
> and
> print('teps.eps')                -> %!PS-Adobe-3.0
>
> No PS -d option is provided.

hmm ... the Mac version of Matlab has supported {"-dps", "-dpsc" , "- 
dps2", "-dpsc2"} for the last few years.

>  Matlab seems not use the suffix to
> decide anything.  Therefore, print('tps.ps')  and print('teps.eps')
> produce identical outputs except for the %%Title: and %%CreationDate.
> However, print -deps 't2eps.eps' and print('t2eps.eps') do not produce
> the same outputs.  The only way to get PS seems to be  
> print('teps.eps').

The following should produce a ps file

	print (gcf, "-dps", "test.ps").

> For PS and EPS the bounding box is "tight" and the same for the
> both print -depsc and print('teps.eps').

I tried the following with Matlab 2008b

 >> plot(1:10)
 >> print(gcf,'-dps','test.ps')

The bounding-box is at the end of the ps-file

%%Trailer
%%BoundingBox:    78   215   548   589
%%Pages: 001
%%EOF

The bounding box is tight. However, the page size is usletter.  To be  
fully compatible we need to render a full page and then crop the  
bounding box. Is it worth the hassle to do that (I don' t know how to  
do that cross-platform).

Regarding pdf output, Matlab does produce a usletter size page. I'm  
much less fluent if the pdf language. While there is no "boundingbox",  
I did notice ...

	MediaBox [0 0 612 792]

Which is 8.5x11 inches. So it appears to me that the changeset is  
doing the correct thing.

In any event, as a result of our discussions I've looked at the  
behavior of Matlab's printing more closely. It appears I've been  
working under false impressions about how the paper* properties are   
intended to be treated. I haven't been paying attention to  
paperpositionmode. When paperpositionmode == "auto" devices directed  
related to printed output (ps & pdf) respect the papersize. The other  
devices render an output that mirrors the image on the screen (I  
think ... I'll need to experiment further to be sure).

Ben