Daniel Arbuckle's Mastering Python
上QQ阅读APP看书,第一时间看更新

Importing a cyclic dependency

There is something that might trip us up when we're importing a module that shares the same package. Sometimes, the module we're importing wants to import us as well. This is called a cyclic dependency. When we try to import a cyclic dependency, we'll almost always get an attribute error exception, as in the following example:

That happens because when we ask Python to import the first module, Python immediately creates a module object for it and begins executing the code in the module.

That's fine, except that, when Python gets to the import statement for the next module in this cycle, it pauses running the code in the first module, leaving it not fully initialized. Even that isn't normally a problem because Python will come back and finish the initialization later.

However, when the second module asks to import the first module, Python just hands it the already allocated, and not fully initialized, module object. When the second module tries to access the variables stored in the first object, many of them will not yet have been created. Hence, an attribute error is raised.