plot() command broken in Octave version 3.0.0 for Red Hat Enterprise?

[Rick Niles]

Your the man!  I had an m-file called "newplot.m" that contained only 
one word: "figure".  That was enough to bring it down.  It works now.  
Sorry about the silliness.  I don't know if this is another personal 
configuration thing but upon exit I get:
error: feval: the symbol `java_exit' is not valid as a function

Finally, I'm running on a very beefy machine it's takes almost 45 
seconds to start octave and about 10 seconds to execute "plot([1 2 3])"

octave:1> tic;plot([1 2 3]);toc
Elapsed time is 7.16495 seconds.
(note after the Elapsed time prints it's still another three seconds 
before anything goes to the plot)

What's up with this slowness?  We have a whole bunch of NFS mounts, 
maybe it's an I/O issue?

Thanks,
Rick.


John W. Eaton wrote:
> On 12-Feb-2008, Rick Niles wrote:
>
> | OK, I've been out of octave development for a while, but something seems 
> | really wrong here:
> | 
> | octave:2> plot([1 2 3])
> | error: can't perform indexing operations for <unknown type> type
> | error: called from `plot' in file `/usr/share/octave/3.0.0/m/plot/plot.m'
> | octave:2> version
> | ans = 3.0.0
> | octave:3>
> | 
> | No matter what type of item I pass to plot() I get the same error.  
> | Could something be corrupted with the RPM?
>
> I can't reproduce this problem with 3.0.0 on a Debian system.
>
> The plot function is fairly short:
>
>   function retval = plot (varargin)
>
>     [h, varargin] = __plt_get_axis_arg__ ("plot", varargin{:});
>
>     oldh = gca ();
>     unwind_protect
>       axes (h);
>       newplot ();
>       tmp = __plt__ ("plot", h, varargin{:});
>     unwind_protect_cleanup
>       axes (oldh);
>     end_unwind_protect
>
>     if (nargout > 0)
>       retval = tmp;
>     endif
>
>   endfunction
>
> Can you track down where the error happens?  Do you have any functions
> in your path called gca, axes, newplot, or __plt__ that are not from
> Octave 3.0.0?
>
> jwe
>