loupgaroublond@xxxxxxxxx wrote: > Hey List, > > I'm working on a couple of projects to get them to use setuptools, but > i've come across an interesting dillema. Some core systems modules, such > as dbus and rpm aren't normally provided as eggs, because they are built > as part of the build process of the original C code. This yields two > issues. > > 1) I cannot use virtualenv to test these packages in a clean room. I > like to test things in as clean of an area as possible, and sometimes > it's difficult to figure out what i've installed might be modifying some > expected behavior. Virtualenv let's you build very light weight > containers for testing python code in, but without distutils, it's hard > to run some setup script or rely on easy_install to provide the needed > bits. > easy_install is of only marginal use when dealing with modules that have to be compiled. Getting built eggs for that case is going to be a hard sell. > 2) How do we normally declare dependencies on these components in > setuptools? If i were to push the package i'm making to the Python > Cheese shop, it would depend on something the user may not actually have > installed. How do i go about declaring these dependencies? > The best thing to do is to not depend on those packages that do not provide egg-info. Trying to describe a dependency on them in the setup.py is doing no one any favors (as your user's cannot satisfy the dependencies so they cannot ever install this.) > To make a short story long, i'm sure there are ways to include code > snippets in dbus, rpm, and possibly other packages that also generate > egg info. This would at least let me declare those packages as > dependencies. The bigger issue i have though, is how can i import them > in some generic fashion into virtualenv? > Now that that's out of the way.... look at pygobject2 for some Makefile rules that may help you. -Toshio -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
