Original report at SourceForge, opened Wed Sep 1 19:07:27 2010

setupext.py is incorrect for Mac OS X: the entry for "darwin" in the "basedir" dictionary is intentionally commented out and should be restored.

As it stands setupext.py must be edited any time one wishes to build matplotlib for Mac OS X.

SourceForge Comments

On Sat Oct 9 20:59:36 2010, efiring wrote:

Russell, I don't know how to handle this; I don't use a Mac. It is not clear to me how to how to raise an exception at an early stage, as you suggest, while still allowing setup to run from make.osx. Maybe John can either come up with an improvement, or close the bug for now.

On Fri Sep 10 12:45:41 2010, reowen wrote:

I have no idea what was wrong that prompted you to disable darwin. All I know is that if I re-enable it and build normally then everything works well. Thus I suggest re-enabling darwin, perhaps with a warning that suggests an alternative in case the build fails.

If you are going to disable darwin I suggest you do it in a way that causes an exception with a sensible message. As it stands what happens is the build fails with mysterious errors, which is painful.

On Sat Sep 4 20:57:59 2010, efiring wrote:

The rationale is given in setup_ext.py:

# it appears builds with darwin are broken because of all the
# different flags the deps can be compile with, so I am pushing
# people to :
#   make -f make.osx fetch deps mpl_build mpl_install

Is this rationale no longer valid?

SourceForge History

• On Sat Oct 9 20:59:36 2010, by efiring: assigned_to: 100