package import dangers

Dave Angel davea at
Wed Oct 7 03:44:35 CEST 2009

Steven D'Aprano wrote:
> On Tue, 06 Oct 2009 18:42:16 +0200, Diez B. Roggisch wrote:
>> The most common problem is that a file is used as module and as
>> executable at the same time.
>> Like this:
>> --- ---
>> class Foo(object):
>>     pass
>> if __name__ == "__main__":
>>    import test
>>    assert Foo is test.Foo
>> ---
>> This will fail when executed from the commandline because the module is
>> known twice - once as "__main__", once as "test".
> Why would a module need to import itself? Surely that's a very rare 
> occurrence -- I think I've used it twice, in 12 years or so. I don't see 
> why you need to disparage the idea of combining modules and scripts in 
> the one file because of one subtle gotcha.
I'm surprised to see you missed this.  A module doesn't generally import 
itself, but it's an easy mistake for a circular dependency to develop 
among modules.  modulea imports moduleb, which imports modulea again.   
This can cause problems in many cases, but two things make it worse.  
One is if an import isn't at the very beginning of the module, and even 
worse is when one of the modules involved is the original script.  You 
end up with two instances of the module, including separate copies of 
the global variables.  Lots of subtle bugs this way.

And there have been many threads right here, probably an average of once 
every two months, where the strange symptoms are ultimately caused by 
exactly this.


More information about the Python-list mailing list