[vtkusers] New vtk debs based on CVS source: some comments.

David Gobbi dgobbi at irus.rri.on.ca
Thu Nov 23 16:48:39 EST 2000


On Fri, 24 Nov 2000, Prabhu Ramachandran wrote:

> Actually for the python bindings all I have is the foll. (debian
> python looks here for python modules) 
> 
> in /usr/lib/python1.5/site-packages/

I hope it checks the current version of python, instead of assuming
that 1.5 is installed.

> <snip>
> drwxr-xr-x    2 root     root         1511 Nov 23 22:31 vtk/
> -rw-r--r--    1 root     root            4 Nov 22 18:50 vtk.pth
> <snip>

The makefiles currently only put the .pth file there, not the
modules themselves.  I'm not sure it makes all that much difference,
so I don't plan to change the makefiles in this regard.
 
> And the stuff in the site-packages/vtk dir is almost all of the stuff
> in vtk_root_dir/python/ excepting the vtkTk* which goes into /usr/lib.
> Works perfectly vtkTkRenderWidget and all.  Besides, I compile the
> stuff in the vtk directory to obtain *.pyo and *.pyc files as well! :)

I wish you were still maintaining the RPMs!  Or even better, if you
could incorporate this functionality into the Makefile.

>     David> Thanks.  After a bit of thought, I decided to make 'dir' on
>     David> the objects to recurse to the subclasses
>     David> (dir(obj.__class__) works the way the old 'dir' did).
> 
> Sorry, I dont get you.  The old bindings did not recurse through so do
> you mean you changed it after 20th Nov. 2000 back to a state where it
> actually does not recurse - which is how the old bindings worked??

No, I mean the initial horde of changes I applied to the wrappers (a few
weeks back) did _not_ include the recursive dir().  I only added the
recursive dir a couple weeks ago.

>     David> The only other change I have planned for the python
>     David> wrappers is better __doc__ strings, both for the methods
>     David> and the classes.  Mainly, the method .__doc__s should
>     David> reflect python calling conventions instead of C++,
> <snip>
> 
> how about the typedef thing Sebastien and you suggested?  Shouldn't it
> make your life easier?

Actually, from the point of view of manufacturing the docstrings, it
doesn't make any difference whether the information comes from the
hints file or from typedefs (but I still think typedefs are a good idea
for other reasons).  

 - David

--
  David Gobbi, MSc                    dgobbi at irus.rri.on.ca
  Advanced Imaging Research Group
  Robarts Research Institute, University of Western Ontario





More information about the vtkusers mailing list