problem building on solaris 10

interpreter. Any suggestions about where to start looking,

    > and what to look for, to debug this will be greatly
    > appreciated.

    > _transforms_module::new_value _transforms_module::new_value
    > _transforms_module::new_point Point::Point
    > _transforms_module::new_value _transforms_module::new_value

   ...snip snip

You have VERBOSE=True in setup.py. While this is fine, it's really
only useful when debugging segfaults in the extension code (not python
exceptions). When debugging python level problems, it suffices to run
with

  > python myscript.py --verbose-helpful
  > python myscript.py --verbose-debug

The default verbose setting is set in your rc file
http://matplotlib.sf.net/matplotlibrc

    > line 9, in ? from backend_agg import FigureCanvasAgg File
    > "/home/erik/util/lib/python2.4/site-packages/matplotlib/backends/backend_agg.py",
    > line 94, in ? from _nc_backend_agg import RendererAgg as
    > _RendererAgg ImportError: No module named _nc_backend_agg

Hmmm.. Do you have Numeric installed? Ie, in the environment in
which you are building, can you do

python

import Numeric

matplotlib supports both Numeric and numarray via the numerix module,
and builds two extensions for each extension code module, ie,
_nc_backend_agg and _na_backend_agg. Which one it builds depends on
what it finds at compile time (Numeric or numarray). If for example
you have only numarray installed, it will build only for numarray.
Unfortunately, the rc file is pretty dumb, and doesn't reflect these
dynamic compile time settings.

But the default rc setting for numerix is Numeric. My guess is that
this is your problem, and can be fixed either by setting numerix :
numarray or installing Numeric and recompiling.

In any case, you will want to turn off VERBOSE in setup.py, flush the
build subdir, and recompile.

Another lurking gotcha along these lines is the backend setting.

  backend : GTKAgg # the default backend
  numerix : Numeric # Numeric or numarray

As with numerix, the backend setting requires you have the appropriate
GUI toolkit installed (eg pygtk) often at compile time if you want to
use a GUI. If you are only doing offline image production you can
make the backend Agg. See also

  http://matplotlib.sf.net/backends.html.

JDH