Issue1143

In a particular project that I work on I have a normal, non-package 
directory that happens to have the same name as a package that I need 
to use from Jython.

It seems that since this directory is in the current working directory 
when I start jython it is masking the real package of the same name.

I imagine this should be easy to reproduce. Let me know if I can help.

I'm actually having a little trouble reproducing this.  What version of
Jython are you using?  Also, could you describe a minimal directory
structure that causes trouble?  I tried an empty "foo" directory with a
sibling foo.py, and importing foo from another .py file worked.  This is
using trunk on OS X.

I just updated SVN and built jython. Still exhibiting the same behavior:

Archimedes ~/code % mkdir foo2
Archimedes ~/code % jython/src/shell/jython
Jython 2.5a3+ (trunk:5523M, Oct 28 2008, 11:33:53) 
[Java HotSpot(TM) Server VM (Sun Microsystems Inc.)] on java1.5.0_15
Type "help", "copyright", "credits" or "license" for more information.
>>> import foo2
>>> foo2.__file__
>>> dir(foo2)
['__name__']
>>> foo2.__name__
'foo2'
>>> 

I would expect "import foo2" to raise an ImportError since there isn't 
any such package.

Doh, this is due to our PathPackageManager attempting to handle Java 
packages that might only contain other packages:

http://fisheye3.atlassian.com/browse/jython/trunk/jython/src/org/python/
core/packagecache/PathPackageManager.java?r=4040#l46

We should make this packageExists() check consider an empty directory to 
not be a package -- but we would have the same issue for an empty 
directory containing another directory. I'm not sure we can handle this 
without some false positives

I wonder if it would be too expensive to check to see if we are looking
at an entire empty directory tree?

I suspect this will get resolved for better or worse when we implement http://www.python.org/dev/peps/pep-0420/ for Jython 3.