Once again, I am having issues with imports...
Until now, I thought the general guidelines were to rarely use 'from x
import y' syntax, except when you really want to copy names over.
However, I have run into issues by following this guideline. So...
1) What is going wrong in the example below?
2) What is a good way to handle imports for packages w/subdirectories?
Here is a sample directory structure:
importtest/
__init__.py
test2/
__init__.py
someclass.py
mytest.py
Here are the definitions of the files:
# importtest/__init__.py
from test2 import *
# importtest/test2/__init__.py
from someclass import *
from test2 import *
# importtest/test2/someclass.py
class SomeClass:
pass
# importtest/test2/mytest.py
import importtest
print importtest.SomeClass
On import I get the following:
>>> import importtest
Traceback (most recent call last):
File "<stdin>", line 1, in ?
File "/home/user/lib/python/importtest/__init__.py", line 1, in ?
from test2 import *
File "/home/user/lib/python/importtest/test2/__init__.py", line 2, in ?
from mytest import *
File "/home/user/lib/python/importtest/test2/mytest.py", line 3, in ?
print importtest.SomeClass
AttributeError: 'module' object has no attribute 'SomeClass'
>>>
The problem seems to be an 'order' issue. importtest/test2/__init__.py
has loaded someclass.py, but it doesn't seem to have copied its contents
into importtest/__init__.py....perhaps it is because it hasn't finished
all of its imports. Is this correct?
So what is a good way to deal with this? In files which contain
implementations, I thought it was best not to use 'from x import y', but
this seems to be the only way to get this to work:
# importtest/test2/mytest.py
from someclass import SomeClass
print SomeClass
Is this the guideline?
Use 'from x import y' for modules within your package.
Use 'import y' for modules outside your package.
Help.
--
http://mail.python.org/mailman/listinfo/python-list
