From owner-help-octave at bevo dot che dot wisc dot edu  Fri Nov 29 10:05:38 1996
Subject: Re: Matlab ginput
From: "petravick at FNAL dot GOV" <petravic@fndaub.fnal.gov>
To: Ted Harding <Ted dot Harding at nessie dot mcc dot ac dot uk>
cc: chelle at bcgn dot grignon dot inra dot fr, help-octave@bevo.che.WISC.EDU
Date: Fri, 29 Nov 1996 10:05:21 -0600 (CST)

We faced this choice a few years ago in a differnt context. We "required"
ginput()-like functionality, and could not choose gnuplot for the same
reasons.

At that time we chose a package called pgplot. IT had the beneift of
optaillying forking off an X handler, so our application did not need an event
loop. This has served us very well, but integrating pgplot introduces it own
blemish, as it is written in f77.

We have compiled it with g77, and I have heard that others have "f2c"ed it,
but it is a problem.

-- DOn


On Fri, 29 Nov 1996, Ted Harding wrote:

> Date: Fri, 29 Nov 1996 15:04:09 +0000 (GMT)
> From: Ted Harding <Ted dot Harding at nessie dot mcc dot ac dot uk>
> To: chelle at bcgn dot grignon dot inra dot fr
> Cc: help-octave at bevo dot che dot wisc dot edu
> Subject: Re: Matlab ginput
> 
> ( Re Message From: chelle at bcgn dot grignon dot inra dot fr )
> > 
> > I think that's impossible, because the use of gnuplot to draw graphics, but,
> > I 'd like to know if it exists an octave function that allows to specify
> > on the graphic window, several points or domain with the mouse and to
> > retrieve this data in an octave object. I think there is a matlab
> > function (~ginput()) that does this.
> 
> There is indeed such a MatLab function (and very useful it can be).
> However, there is no such function with octave-1.1.1.
> 
> Nor, so long as graphics depends on gnuplot, is there likely to be: the
> gnuplot display does its work independently of octave once it has received
> the input from octave: only gnuplot knows where on the screen the
> coordinates are plotted, and gnuplot won't tell. You can't use the mouse
> to ask gnuplot for the coordinates of points it has plotted.
> 
> However, if a different graphics package were used (which one, anyone?),
> which had the capability of outputting the coordinates of points selected
> by the mouse, then it would be possible to write such an octave function.
> Either the coordinates could be written to a file which could be read by
> octave. or maybe pipes or sockets could be used.
> 
> The statistical software package xlispstat is very good at graphics, and
> allows this sort of selection very flexibly. I have been toying with the
> idea of interfacing octave to it (but would have to grapple with LISP).
> 
> Another possible benefit that could result from such a step would be the
> possibility of dynamic graphing -- adding points/lines to a plot in real
> time as they are computed.
> 
> Has anyone else any ideas along these lines, or has gone further into
> xlispstat or similar than I have?
> 
> Best wishes to all,
> Ted.                                    (Ted dot Harding at nessie dot mcc dot ac dot uk)
>