Actually, I found a (pretty ugly) workaround to my problem: In every __init__.py of the subdirectories I created, I add the following lines:
import os.path,sys
for directory,sub_dir,files in os.walk(os.path.normpath(os.path.join(os.path.dirname(__file__),".."))):
sys.path.append(directory)
That add all of the sub-directories of the parent directory to the sys.path. I can then move my files in the directories I want without having to change a single line in them.
it may not be very elegant, but for the moment, it will do.
Actually, after thinking about it, I think that my problem is more a problem with the os than with python. subdirectories are the only method most OS provide for file organization. If there was a method for, for example, add labels to files and then browse files according to the labels, I wouldn't have this problem. (ok, I could arrange that with the names of the files, but it doesn't feel the same for me).
Jason> Actually, unless I am wrong, you don't need to add the path of the package with a .pth . It should be done automatically when you import it.
Giovanni>
>The usual way is to do "import pkgN" and then dereference
What do you call 'dereference'? do you mean something like 'import spam.foo foo=spam.foo' ?
>My preferred way is to have __init__.py just do "from submodules import *", and then each submodule defines __all__ to specify which are its public symbols.
Your solution soud interesting (although it doesn't exactly do what I want in the present case). I would consider it when I have to do hierarchical packages. However, I think I read somewhere that "from module import *" statements could cause (probably small) performance issue. Do you think this can be the case here?
Anyway, thank you all for your very interesting replies. (and sorry for the double-post)
TokiDoki
-- http://mail.python.org/mailman/listinfo/python-list
