Re: 3.6.1 produces eps files that are unusable on Debian wheezy

[Daniel J Sebald]

On 03/04/2012 03:10 PM, Mike Miller wrote:
> On Sun, Mar 4, 2012 at 12:58 AM, Daniel J Sebald<address@hidden>  wrote:
> > On 03/03/2012 10:09 PM, Mike Miller wrote:
> > > On Sat, Mar 3, 2012 at 10:57 PM, Dmitri A. Sergatskov
> > > <address@hidden>    wrote:
> > > > BTW, the problem with "{}" exist with the original doc files in octave
> > > > tar ball.
> > > > E.g. in doc/interpreter
> > > > gs mesh.eps
> >
> > [snip]
> >
> > > > > > showpage, press<return>    to continue<<
> > > >
> > > >
> > > > I am not sure if we should consider this a bug.
> > >
> > >
> > > Thanks for finding that, at least it's consistent.  I'm still not
> > > convinced whether the problem is with octave, gnuplot, or gs.  I'll
> > > poke around some more.
> >
> > I consider it a bug somewhere along the way.  The root issue might be
> > confusion arising from gnuplot's dual method of specifying fonts, but I
> > believe it is Octave where the simplest fix is.  If there is an issue with
> > gnuplot inconsistencies, let me know; I'll argue that gnuplot should have
> > font specification consistency.
> >
> > Mike, have you tried adding this line to your startup scripts?
> >
> >
> > set(0, 'defaultaxesfontname', 'Helvetica')
>
> That works once I have octave built and installed, but doesn't help
> when running with -f, such as in building the doc directory.  Your
> other suggestion to use a single space as the font name gave the same
> error.
>
> I have found a resolution for why my ghostscript is failing with the
> same file that works for everyone else.  Here is the gs command that
> finally works for me with the original eps files:
>
> gs -sDEVICE=x11 -dNONATIVEFONTMAP<file>
>
> I used the -dDEBUG switch and it seems like ghostscript is possibly
> failing on my system because of too many installed fonts.  It keeps
> printing one font name after another until I see the same "/limitcheck
> in /findfont" message.  It never gets through all the available fonts
> to get to the part where you guys are seeing "Substituting font
> Courier for {}."

My guess would be that Debian uses different compile switches and 
options than others.

You say you aren't seeing "Substituting font Courier for {}."  Is 
ghostscript failing on the search for {}?  Perhaps since it is a rather 
meaningless font, ghostscript is searching and searching for a suitable 
font.


> To me it seems like gnuplot is working correctly, I'd still like to
> know the reason for the "{}" font name in octave in the first place,
> but the real problem for me is back to ghostscript.  If someone sent
> me a postscript file using a font that I don't have, I would get the
> same error.
>
> Thanks for the help... still can't build from hg because of this though :)

Why can't you build?  Is there some part of the process that creates an 
EPS file and tests whether it is valid?  Is there a build option to 
disable such a test?  One thing you could do after unpacking and before 
building is touch up your file so that

  if (isempty (t.fontname) || strcmp (t.fontname, "*"))
    fnt = "{}";
  else
    fnt = t.fontname;
  endif

looks like

  if (isempty (t.fontname) || strcmp (t.fontname, "*"))
    fnt = "";
  else
    fnt = t.fontname;
  endif

Dan