[vtkusers] vtkChartXY with inverted vtkAxis labels

[Marcus D. Hanwell]

On Tue, Feb 1, 2011 at 1:07 PM, Joseph Melnyk <JMelnyk at pile.com> wrote:
> I had previously tried the nightly version but ran into other issues with charts.  Besides those, the new release does address the need for inverted axes; so thanks much for that!

If there are bugs you can report them in our bug tracker, and I will
of course look into them.
>
> Regarding the other chart issues, I've been looking into those further and their behavior is quite peculiar.  For example, setting the color of the axis title makes the legend go away (even if it's explicitly set to be shown), setting an axis title at all results in it not being shown, setting axis behavior to FIXED results in the tooltips not appearing, etc.  Some of these seem to exist only in my code (which is quite simple); others also appear when I place them into the tests.  Am I missing something here?  Is it perhaps because I'm using them under C++/CLI, with the render window bound to a Windows Form PictureBox?

I have never used used PictureBox and am unfamiliar with CLI (other
than to mean command line interface). I have not seen these bugs, and
you can look at ParaView as an example of an application where these
properties are manipulated and tested on Windows, Mac and Linux
operating systems.
>
> On a related note, is there an easy way to swap axes?  For example, if I want to have the rows of my table correspond to Y values rather than X values.  I had accomplished this before by swapping column indeces (i.e. vtkPlot->SetInput(table, YColIndex, XColIndex)).  This has the side effect of setting all names in the legend to the X column's index, however.  So I was explicitly setting them then with vtkPlot->SetLabel; but this function doesn't appear to work any longer.  Using it at all results in the legend name being blank.  I could replicate this issue with the tests using C++/CLI and binding the render window to a Windows Form PictureBox.  But using Win32, there's no issue.

It sounds like something strange is happening, these functions are
well tested in VTK and ParaView and I have not seen any regressions.
>
> So perhaps this is related to use of C++/CLI after all.  I had none of these issues with the 5.6.1 release, fwiw.

It could be, I am not familiar with your development environment. It
sounds like something pretty bad is happening for things to keep going
blank - perhaps a bad build? We test routinely using Visual Studio
2008 and 2010 on Windows, and the tests seem to pass there without
issue.
>
> Thanks again for your help!
>
It's OK, if there is anything that can be reproduced with our C++
tests I can take a look, but I am not aware of Windows issues at this
time (other than a few antialiasing artifacts.

Marcus