[changeset] printing multipage postscript files

[Ben Abbott]

On Jun 3, 2009, at 11:16 AM, Petr Mikulik wrote:

>>>> Please check that this works are you expect. I've tested the  
>>>> three examples
>>>> below on Mac OSX with no problems.
>>>>
>>>> surf(peaks)
>>>> print -dps append.ps
>>>> sombrero (25)
>>>> print -dps -append append.ps
>>>>
>>>> surf(peaks)
>>>> print -dpdf append.pdf
>>>> sombrero (25)
>>>> print -dpdf -append append.pdf
>>>>
>>>> surf(peaks)
>>>> print -dps append.ps
>>>> rename ("appdend.ps", "append.pdf");
>>>> sombrero (25)
>>>> print -dpdf -append append.pdf
>>>>
>>>> The last one I hadn't expected to work, but ghostscript handled  
>>>> it nicely.
>>>
>>> With postscript, it works well. (The option is really "-append".)
>>>
>>> However, appending to pdf fails on my system:
>>>
>>> x=1:100;
>>> plot (x,x)
>>> print -dpdf z.pdf
>>> plot (x,-x)
>>> print -dpdf z.pdf -append
>>> plot (x,x.*x, x,x+20)
>>> print -dpdf z.pdf -append
>>>
>>> However
>>> gs -dNOPAUSE -sDEVICE=pswrite -sOutputFile=new.pdf \
>>> z1.pdf z2.pdf -c quit
>>> works OK, so there seems to be a some small mistake in the patch.  
>>> Can you
>>> please fix it?
>>
>> The example you gave works for me. If you can isolate the problem,  
>> I can patch
>> Octave's sources.
>
>
> My gnuplot supports "set term pdf" (it is compiled with "pdfcairo"
> terminal). However, it seems that octave goes via postscript file  
> instead --
> I've put some printing commands to print.m, see below. The failure  
> happens
> because the original pdf file (the new plot to be appended to, i.e.  
> the
> original z.pdf) is called below
> 	/tmp/oct-F79Jrl.ps
> but it does not exist. The new plot
> 	/tmp/oct-byjhyV.ps
> exists (but it is postscript, not pdf!).
>
> Does this help?

ahh! ok, I'll take a look at fixing this tomorrow.

Ben