Re: File based terminal behaviour

[Ethan M. <merritt@u.washington.edu>]

On Wednesday 17 August 2005 05:21 am, Gareth Wilson wrote:
> 
> In an interactive terminal (x11 in our case) this is of no consequence but when
> attempting to save the output to a file (I have used both png and SVG) the
> ordering is critical as the first plot command encountered outputs to the file,
> not allowing subsequent replots, clearing, adding of titles etc.

If you are happy with the x11 output, then I would hazard a guess that your
best choice may be to bind a hot-key for x11 that issues a replot command to the
file-based terminal of your choice:
	bind "ctrl-p" "set term png; set output 'myplot.png'; replot; set term x11"

Then you can adjust the plot to your liking in an x window, and hit <control>p
to save it to a file.

For production work you'd need a more complicated binding that incremented the
file name somehow.
 
> This behaviour is not encountered by multiplots which seem to be constructed in
> a temporary terminal and then output once completed (signified by the unset
> multiplot). I cannot get this behaviour to be replicated for single plots 
> however.

I'm afriad I don't recognize the symptoms you describe here.  There is no
temporary terminal involved in multiplot.

 
> From reading the postings here and on Soureforge I gather that the png terminal
> is usually piped to Imagemagick's display

Well, it's the way *I* usually do it, at least if I'm sitting at the terminal.
But primarily I use the png terminal in automated scripts at the back end of
a web form, in which case they are not piped to ImageMagick but are saved as
temporary files and linked back into a dynamically generated web page.

> 
> Would it not therefore make sense to have the option that once a user enters a 
> 'set term png', 'set output foo.png' to construct the plot as multiplot does,
> and then output the completed plot once the 'set output' command closes the 
> file?

That is, in fact, the way they are normally used.
Could you try to describe the probably you are having in more detail?
I can't understand what multiplot has to do with anything.

-- 
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195