prabhu_r at users.sf.net
Sat Jul 16 12:28:46 CDT 2005
>>>>> "Perry" == Perry Greenfield <perry at stsci.edu> writes:
Perry> Since John Hunter is away I'll try to answer that as best
Perry> as I can. I think the current thinking on 3D capabilities
Perry> is that it may be worth adding limited capabilities such as
Perry> surface plots and perhaps 3-d scatter plots. I doubt that
Perry> there will be any attempt to try to duplicate any
Perry> significant fraction of what VTK provides and equivalent.
Perry> The motivation for including any 3-d capabilities is to
Perry> avoid users having to install VTK (which can be daunting
Perry> for some). Also, I think ultimately it is desired that
Perry> matplotlib optionally support VTK plotting within its
Perry> figures (i.e., it won't be necessary to install VTK unless
Perry> you want those capabilities).
I sincerely hope that we don't end up with 10 different 3d packages,
each doing its thing slightly different.
I've worked quite hard on tvtk and the core mayavi2 infrastructure.
I've been trying to use a lot of Enthought's open source tools to do
this job right by focussing on the "model" and less on the view. Here
is more information:
I think a more constructive thing to do would be to make prebuilt VTK
binaries available. Debian ships with VTK (under 11 architectures!),
VTK-4.4 binaries are available for Win32 via Enthon, IIRC VTK rpms are
also available. If this is not enough, surely it would be easier for
folks to just contribute a build for a favorite platform every year or
6 months. It is not too hard to do if you are used to building
software. But it is work and not a trivial thing either.
More information about the SciPy-user