gnuplot development

Dieter Jurzitza wrote:

> Converting gnuplot output from ps to pdf (set terminal postscript 
> enhanced ...) via ghostscript (either ps2pdf or better epstopdf) I found   
> that the default bounding box in XXX.ps out of gnuplot, XXX.ps being the 
> output causes somewhat too large distances between the figure and the 
> caption /and / or the preceeding text in TeX (matter of taste, though ...). 

Careful here.  "set term post enhanced" is not meant to give you any 
usable bounding box at all.  If you want a meaningful bounding box, you 
really should use the "eps" option.

> functionality. This script implies the usage of epstopdf, it does not work 
> with ps2pdf. You need both sed and ghostscript installed.

That seems overly complicated --- ghostscript already comes with a 
script that automatically calculates a "perfect" bounding box.  It's 
called ps2epsi, and I'm almost completely sure that other tools already 
exist which do similar work.  I.e. you could just ps2epsi | ps2pdf

OTOH, if what you want to produce is a PDF, why not just use the PDF 
terminal (assuming you're in a position to compile it in)?

Dear Mr. Br=F6ker,
dear listmembers,
well, I have been on the webpage of the pdflib folks, and, well, they want =
to=20
charge something like 450 Euros per license if I got it right - this is qui=
te=20
expensive for a library the advantage of which I cannot see regarding this=
=20
specific issue. I am perfectly satisfied with the quality of the output I g=
et =20
the way I am using it right now. I will test your suggestions and will be=20
happy if things turn out simpler than I have them right now.
IMHO gnuplot has made several very important steps forward regarding the=20
flexibility of the output. Thanks for all the ongoing work!

Take care & thanks for the pointer



Dieter Jurzitza


Am Donnerstag, 21. Juli 2005 15:40 schrieb Hans-Bernhard Broeker:
*****
>
> OTOH, if what you want to produce is a PDF, why not just use the PDF
> terminal (assuming you're in a position to compile it in)?
>
*****

=2D-=20
=2D----------------------------------------------------------

                               |
                                \
                 /\_/\           |
                | ~x~ |/-----\   /
                 \   /-       \_/
  ^^__   _        /  _  ____   /
 <=B0=B0__ \- \_/     |  |/    |  |
  ||  ||         _| _|    _| _|

if you really want to see the pictures above - use some font
with constant spacing like courier! :-)
=2D----------------------------------------------------------

Dear Mr. Br=F6ker,
dear listmembers,
test on your own. Using dummy.plt as I have been providing with my last ema=
il=20
but replacing "set terminal postscript enhanced landscape" by "set terminal=
=20
postscript eps" using gnuplot-4.1.0:

gs -sDEVICE=3Dbbox ...
gives me a bounding box of 48 39 399 293 for dummy.ps, (1)

the orignial file dummy.ps created by gnuplot contains a bounding box=20
information of 50 50 410 302 (2)

using epstopdf on the file containing BoundingBox (1) results in an=20
appropriate picture, all arrows remain visible in the pdfoutput.

In contrast, applying ps2epsi followed by ps2pdf on the output file contain=
ing=20
the BoundingBox (2) results in arrows cut out of the visible range.

Calling gs with sDEVICE=3Dbbox actually does not care for any given=20
bbox-information in the file but computes them from the raw ps data - and=20
those are correct. Looking into the actual pdf-files shows that the MediaBo=
x=20
information differs in both cases (what is expected due to the differences =
in=20
appearance).

But maybe I did another error, so, if you find the time, I appreciate if yo=
u=20
can recheck.

The way I suggested may be complex - but it works.

Take care



Dieter Jurzitza

Am Donnerstag, 21. Juli 2005 15:40 schrieb Hans-Bernhard Broeker:
> Dieter Jurzitza wrote:
> Careful here.  "set term post enhanced" is not meant to give you any
As mentioned above. gs -sDEVICE=3Dbbox does not care for the actual Boundin=
gBox.=20
It computes a new one based on the real graph data.
> That seems overly complicated --- ghostscript already comes with a
> script that automatically calculates a "perfect" bounding box.  It's
> called ps2epsi, and I'm almost completely sure that other tools already
> exist which do similar work.  I.e. you could just ps2epsi | ps2pdf
See commentary above

=2D-=20
=2D----------------------------------------------------------

                               |
                                \
                 /\_/\           |
                | ~x~ |/-----\   /
                 \   /-       \_/
  ^^__   _        /  _  ____   /
 <=B0=B0__ \- \_/     |  |/    |  |
  ||  ||         _| _|    _| _|

if you really want to see the pictures above - use some font
with constant spacing like courier! :-)
=2D----------------------------------------------------------