Re: problem with 3d graph on win32

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

On Wednesday 18 March 2009 14:25:52 Allin Cottrell wrote:
> Every now and then I build a binary of CVS gnuplot for win32
> (cross-building on Linux).  The last time I did that was several
> months ago, and the binary worked fine.  I recently (2009-03-02)
> updated and re-built.  The new binary seemed OK at first, but now
> I notice that it crashes whenever I try to use the "handle"
> cursor to rotate a 3d plot in the windows terminal.

I see no Windows-specific changes since 7-Nov-2008, and that bunch
was supposed to be conditional on the configuration option 
WGP_CONSOLE.    It would help to know whether your previous
working version was built before or after that date.

In fact, in all of 2008 I only see about 6 windows-specific 
changes in ChangeLog, and most of those are both trivial and not
specific to 3D plots. 

As to recent changes that would affect 3D in general, I see

2009-03-11  Ethan A Merritt  <merritt@u.washington.edu>
        * src/graph3d.c (do_3dplot): Define the clipping area in 3D plots to lie
        between the left-most and right-most graph box edges. This is a change!
        The intent is to allow the canvas terminal to use the plot's BoundingBox
        as a zoom region.  If it causes problems for other terminals, we will
        need to create and use a separate BoundingBox for this purpose.

2009-01-07  Ethan Merritt  <merritt@u.washington.edu>
        * src/util3d.c (edge3d_intersect):  Fix long-standing bug in which
        the z coordinate was clipping against xmax rather than zmax.

Beyond that, the changes I notice mostly depend on the specific type of
plot.  Could you check whether the problem is only present for a simple
line plot?  +/- pm3d?  With all axis labels and tick marks turned off?

Oh, one other annoying possibility...
Does it only happen when the executable is run under Windows?
I.e., if you run it in linux under Wine, does it still crash?


> Debugging on Windows is not very convenient for me so I haven't
> tried that yet.  Does anyone happen to have an idea of what might
> be wrong here?  I've briefly scanned CVS, but a lot has gone on
> lately so I couldn't isolate anything "suspicious".  Thanks.
> 
> -- 
> Allin Cottrell
> Department of Economics
> Wake Forest University
> 
> 

-- 
Ethan A Merritt