On Sun, 7 Dec 2008, Eric Faurot wrote: > For python modules not using MODPY_SETUPTOOLS, the installed .pyc > files refer to their source .py files in the fake directory instead of > LOCALBASE. I am not sure how bad it is in general, but this is at > least a problem for some py-twisted regress tests which fails because > of this. > > I see no reason why it should not be handled correctly as in the > "MODPY_SETUPTOOLS=Yes" case.
Have you tested this with a good cross section of py-* packages? I'm unsure whether this needs a crank for all the non-setuptools packages. It does mean a material change in what is packaged, right? -d