gnuplot terminal, "postscript", is not available

[Ben Abbott]

On Mar 23, 2009, at 8:12 PM, Ivan Sutoris wrote:

> On Mon, Mar 23, 2009 at 11:25 PM, Ben Abbott <bpabbott at mac.com> wrote:
>>
>> On Mar 23, 2009, at 11:12 AM, Ben Abbott wrote:
>>
>>>
>>> On Monday, March 23, 2009, at 09:29AM, "Ivan Sutoris"
>>> <ivan.sutoris at gmail.com> wrote:
>>>>
>>>> Bug report for Octave 3.1.54+ configured for i686-pc-linux-gnu
>>>>
>>>> Description:
>>>> -----------
>>>>
>>>> After compiling recent tip (9006:979d0492ff40), I can't print  
>>>> figures
>>>> to files (it worked with devel. version before and also in  
>>>> 3.0.3). I
>>>> get "error: gnuplot_drawnow: the gnuplot terminal, "postscript", is
>>>> not available." I am using gnuplot 4.2 and the problem seems to  
>>>> occur
>>>> in line 44 of gnuplot_drawnow.m:
>>>>
>>>> available_terminals = __gnuplot_get_var__ (plot_stream,
>>>> "GPVAL_TERMINALS");
>>>>
>>>> After this line, available_terminals contains NaN, so no terminal  
>>>> is
>>>> recognised. Could this be caused by this version of gnuplot not
>>>> working corectly with __gnuplot_get_var__?
>>>>
>>>> Repeat-By:
>>>> ---------
>>>>
>>>> plot(randn(100,1))
>>>> print myfig.eps -depsc
>>>>
>>>
>>> Ivan,
>>>
>>> My mistake. The GPVAL_TERMINALS variable apparently is only  
>>> available for
>>> gnuplot's developer's sources.
>>>
>>> I'll provide a patch later. In the meantime, you can short-circuit  
>>> the
>>> check in gnuplot_drawnow by ...
>>>
>>>     plot_stream = open_gnuplot_stream (2, []);
>>>     available_terminals = __gnuplot_get_var__ (plot_stream,
>>> "GPVAL_TERMINALS");
>>>     available_terminals = regexp (available_terminals, "\\b\\w+\\b",
>>> "match");
>>> -     if (any (strcmpi (available_terminals, gnuplot_trim_term  
>>> (term))))
>>> +     if (true)
>>>
>>> Ben
>>
>> I've pushed the patch.
>>
>> Ben
>
> With the patch it works again, thank you for quick solution :)
> However, when printing I get many (dozens) warnings that probably
> should not be there:
>
> warning: implicit conversion from real matrix to real scalar
>
> I tried to identify the problem using debug_on_warning and it seems
> that these occur in __go_draw_figure__.m and __go_draw_axes__.m with
> fprintf and fputs commands, for example line 67 in the first one:
>
> fputs (plot_stream, "\nreset;\n");
>
> Here fputs expects just one file identifier, but actually plot_stream
> is two element vector and so the conversion (and warning) takes place.
>
> Regards
> Ivan Sutoris

Thanks Ivan,

I've been intending to track down that warning, but hadn't been  
successful.

Ben