[Distutils] [setuptools] Bug in setuptools ImpLoader

[Jeremy Kloth]

On Friday 08 September 2006 1:16 pm, Phillip J. Eby wrote:
> At 12:24 PM 9/8/2006 -0600, Jeremy Kloth wrote:
> > >From PEP 302:
> >
> >   The load_module() method has a few responsibilities that it must
> >     fulfill *before* it runs any code:
> >
> >     - If there is an existing module object named 'fullname' in
> >       sys.modules, the loader *must* use that existing module.
> >       (Otherwise, the reload() builtin will not work correctly.)
> >       If a module named 'fullname' does not exist in sys.modules,
> >       the loader must create a new module object and add it to
> >       sys.modules.
> >(emphasis on must)
> >
> >The current implementation always reloads the module if it exists.
>
> Yes, that's the *intent* of the above paragraph.  An importer that
> *doesn't* reload under these circumstances will break the reload() builtin.

I read the paragraph as stating that the loader *returns* the module from 
sys.modules instantly or it would break reload().  After re-reading PEP 302 
several times, it becomes clear that the intention is that the module's 
contents are replaced before returning the module.

> >   This
> >problem exists in both the 0.6 and 0.7 series.  Since pkgutil is in Python
> >2.5, this is an issue there as well.
>
> Unless I've misunderstood you, this is not a problem, it's the correct and
> intended behavior.  When a loader is asked to load a module that's in
> sys.modules, it's supposed to *reload* it.  Otherwise, the reload() builtin
> won't work.

It would be worth adding a sentence stating that loader.load_module() is the 
same as a reload() if the module is already imported (just like the docs for 
imp.load_module).

My confusion, no problem here except for in the code that was expecting my 
concept for things to work.

-- 
Jeremy Kloth
http://4suite.org/