Re: [changeset] suppress unexpected shift in eps image by gnuplot

[Ben Abbott]

On Mar 3, 2009, at 12:57 AM, Ben Abbott wrote:

> I found a problem with eps output (via gnuplot), and pushed a  
> changeset that empirically suppresses the problem.
>
> changeset:   8901:821f0242e8c1
> tag:         tip
> user:        Ben Abbott <address@hidden>
> date:        Tue Mar 03 00:45:00 2009 -0500
> summary:     gnuplot_drawnow.m: Fix unintended shift of plot image  
> for the gnuplot postscript+eps terminal.
>
> The eps image was shifted unexpectedly (left margin larger and right  
> margin smaller).
>
> A bug report has been filed for gnuplot.
>
> https://sourceforge.net/tracker2/?func=detail&aid=2656342&group_id=2055&atid=102055
>
> I've verified this works with gnuplot 4.3.x (present developers  
> sources) and 4.2.3.
>
> Ben

After discussing the problem with the gnuplot developers, it was  
concluded that the problem lies in the interpretation of the image.  
Specifically, lower left corner of the image designated by (0,0) or by  
the bounding box?

As best I can tell, ghostscript (gs) favors (0,0) and ghostview (gv)  
favors the LL corner of the bounding box.

Thus for postscript output (which is generally intended to be printed)  
this changeset favors the gs perspective. For eps files the changeset  
favors the gv perspective.

Before committing this changeset (which essentially reverts my prior  
one), I thought I'd give others an opportunity to comment.

Ben

changeset-epsfile.patch
Description: Binary data