[SciPy-user] Clarification

Joe Cooper joe at enthought.com
Sat Jul 16 20:28:10 CDT 2005

If it is any consolation, we all have trouble building and maintaining 
VTK installations.  ;-)

Further consolation may be found in the knowledge that MayaVi and 
MotPlotLib and all dependencies are included in all newer Enthought 
Edition Python releases (like the current test release on the 
website...not that I'm recommending you use that one, as it has some 
pretty ugly warts), and Prabhu has been very helpful in making sure we 
include all of the right bits so that MayaVi and VTK work correctly out 
of the box.  I will post a new Enthought Edition official release with 
the latest MayaVi release and VTK 4.4 next week sometime--it reverts a 
lot of components back to the older versions found in the 1057 build, 
due to incompatibilities and instabilities and other bad things, but the 
plotting stuff is reasonably cutting edge.  These things take time, 
because we have competing pressures for extreme stability of both 
software and interfaces on one side and for rapid inclusion of the 
latest versions of everything on the other.  The bundle is just so 
bloody huge, it's a crap shoot to upgrade even a single element.  There 
seems to finally be some movement within the python community for easier 
to install and update software bundles, ala CPAN for Perl, such that it 
may soon be easy for folks to safely upgrade some components 
individually.  That will be nice.

Alan G Isaac wrote:
> On Sat, 16 Jul 2005, Prabhu Ramachandran apparently wrote: 
>>I think a more constructive thing to do would be to make 
>>prebuilt VTK binaries available. 
> I am a relatively new SciPy user. 
> Here was my experience. 
> 1. After some messing around, learn that I should use 
>    Matplotlib rather than *plt. 
> 2. Learn that Matplotlib does not support 3D surface plots. 
> 3. Learn that people are recommending MayaVi. 
> 4. Go to MayaVi website and give thanks that there is 
>    a Win32 binary that "comes bundled with everything 
>    necessary to run MayaVi."  (Of course this means that 
>    I read past the "recommendation" to install from sources 
>    ...) 
> 5. Install the binary and discover, alas, that I did not get 
>    the pyvtk module with it. And test_vtk.py wants in 
>    addition a vtkpython module. Then I find this:
>         "If you have installed MayaVi from the sources and 
>         are not using a binary release, then you can use 
>         MayaVi as a Python module."
>    Aha.  That is what that "recommendation" was about.  Can 
>    we agree that "recommendation" at this point appears 
>    a bit understated? 
> 6. Find nothing in the docs about how to add the module(s) 
>    to work with my MayaVi installation, except of course the 
>    recommendation to install from source.  (Of course, 
>    despite the sanguine assessment on the MayaVi page, 
>    I recall the various warnings of others, plus I now feel 
>    "once burned".)
> 7. Give up and keep using gnuplot for now. 
> If this sounds like a complaint, it is not.
> It is meant to be informational.

More information about the SciPy-user mailing list