> I still think it is unfortunate that we are starting to extend pip to > be a tool for developers to create distributions. It would be better > of pip was kept as an install tool, and we added the utilities for > creating distributions separate. >
I understand where you're coming from, but a few thoughts: 1) pip is *currently* very much a build tool in that it build/installs from source archives, but I understand the new model is for pip to eventually be working with pre-built wheels much of the time, with no build system required. 2) the motivation for "pip wheel" is *not* really for building single wheels for the project you're developing. For that use case, I agree it makes more sense conceptually to install the "wheel" package and use it's "bdist_wheel" setuptools extension (i.e. "python setup.py bdist_wheel" ) 3) the real motivation for "pip wheel" (which is a builder convenience tool) is to help people *install* from wheels *now* given that pypi won't be full of wheels for a time to come. This allows people to quickly and easily get all their source archive dependencies converted with a single command (if you're using requirements files) and start gaining the benefits of wheel with very little fiddling. 4) even when pypi is full of wheels, I can imagine people wanting to build certain dependencies with different build options, and "pip wheel" could help with that. 5) I can imagine "pip wheel" disappearing at some point down the road. We'll have to see, but until then, "pip wheel" is going to very critical IMO in getting people familiar with and actually using wheel-based installs. > The other option is of course that we start adding all sorts of > development commands to pip, such as build, test, sdist etc. But I do > think it's the wrong place. > I've thought of that too, but that's a discussion for another day or parallel universe. Marcus
_______________________________________________ Distutils-SIG maillist - Distutils-SIG@python.org http://mail.python.org/mailman/listinfo/distutils-sig
