missing pdf terminal; was --> [changeset] print.m (matlab compatibility)

[Ben Abbott]

On Mar 10, 2009, at 5:35 AM, John W. Eaton wrote:

> | set (0, "defaultaxesfontsize", 12)
> | set (0, "defaultetxtfontsize", 12)
> | set (0, "defaultaxesfontname", "*")
> | set (0, "defaultetxtfontname", "*")
> | figure (1)
> | clf
> | surfl (peaks ())
> | xlabel ("X")
> | ylabel ("Y")
> | zlabel ("Z")
> | title ("surfl (peaks ())")
> | set (gcf, "paperunits", "inches")
> | set (gcf, "papersize", [6, 4.5])
> | set (gcf, "papertype", "<custom>")
> | set (gcf, "paperposition", [0, 0, 6, 4.5])
> | print (gcf, "-deps", "test_surfl.eps")
> | print (gcf, "-dpdf", "test_surfl.pdf")
> | print (gcf, "-dpng", "test_surfl.png")
> | print (gcf, "-dpng", "-r300", "test_surfl-300dpi.png")
>
> When I try these commands, the pdf version fails because my gnuplot
> (4.2.4, Debian testing) wasn't built with pdf support (why not?) and
> then gnuplot spews a lot of "invalid character" messages.  Maybe we
> need to detect whether a terminal is supported and then fail a bit
> more gracefully.  But don't worry about fixing that in this changeset.


Some weeks ago I wrote a short function to obtain values for   
gnuplot's public variables.

... which can be used to obtain the version and patch-level

	ver = __gnuplot_get_var__ (gcf, "GPVAL_VERSION");
	level = __gnuplot_get_var__ (gcf, "GPVAL_PATCHLEVEL");

... the X11 window id

	window_id = __gnuplot_get_var__ (gcf, "GPVAL_TERM_WINDOWID");

... as well as the available gnuplot terminals

	terms = regexp (__gnuplot_get_var__ (gcf, "GPVAL_TERMINALS"), "\\b\\w+ 
\\b", "match");

Shall I commit this function and add a check to gnuplot_drawnow to  
verify the terminal type is available in gnuplot just after the plot  
stream is opened?

Ben