Python 2.3.2 Windows platform

Python is unable to import modules in ZIP files created 
by the PKZIP command line tool, version 2.50.

Logged In: YES 
user_id=21627

Can you please attach a small sample ZIP file that
demonstrates the problem? I don't have the PKZIP command
line tool.

Logged In: YES 
user_id=700058

More information.

It would appear that the problem is not PKZIP's command line 
archiver as it also happens with ZIP archives created with the 
zipfile module.  Rather the problem has something to do with 
packages in ZIP file.  To reproduce this problem:

1.  Create a sub-directory under c:\Python23 (or whereever 
you installed Python) named test.
2.  Unzip the files from the ZIP archive into the directory.
3.  Start the command line interpretor and enter:

from test import testing

Should display "test"

4.  Exit the command line interpretor.
5.  Delete all *.pyc files from the test directory.
6.  Restart the interpretor and enter the following:

import zipfile
z = zipfile.PyZipFile ('test2.zip', 'w', zipfile.ZIP_DEFLATED)
z.writepy ('test')
z.close()

It should have created a test2.zip archive.

7.  Exit the interpretor.
8.  Verify that test2.zip contains two pyc files inside the test 
folder.
9.  Restart the interpretor and enter the following:

import sys
sys.path.append ('c:\\Python23\\test2.zip')
from test import testing

You should get:
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
ImportError: cannot import name testing

Now, if you created a ZIP archive, with the PY files, it imports 
fine.

Logged In: YES 
user_id=700058

Changed summary to better reflect what the problem is.

I just tried out this test with the trunk and 2.5.2 from svn (on OS X,
not sure if that might make a difference), and didn't get the reported
error.  Since the issue has apparently been fixed at some point, is
there any reason not to close this issue?

I had forgotten about this bug.  I haven't been able to test this on 
2.5 as I have not worked with Python for a while.  I think if you are 
satisfied that the problem doesn't exist, then close it.  If I find it 
again, I can certainly log the problem again.  BTW, did you manage to 
test whether the problem existed in the version that I reported it in?

I just tried it on 2.5 and 2.6, since I didn't have the earlier versions
available.  If I find some time I'll see if I can try it on 2.3 & 2.4;
maybe that will shed some light on whether it's been fixed or if is
something specific to Windows that could still be broken.

It seems most people agree that this "works for me".