Suggest closing this one without merging. Project is building fine from setup.py without these changes and there is at least one other Makefile problem. Needs someone who has a use for the Makefiles to investigate.

i'd like to get rid of the Makefile altogether, if it's not useful. this can (and should) all be replaced by setup.py stuff, IMHO. yet this probably doesn't block py3 porting, so keeping this open until someone deals with the Makefile (or opens a more targeted issue to do so).

This one also for the record:

Called by top-level make locale.

Suggest closing this one without merging. Project is building fine from setup.py without these changes and there is at least one other Makefile problem. Needs someone who has a use for the Makefiles to investigate.

I like Makefiles, but I don't have the energy to investigate this. +1 for closing (or even removing the Makefile entirely).

(IMHO in a Python project Makefiles are primarily a tool for maintainers. They make sense when the maintainer finds them useful. I find them useful: all of my personal Python projects have a Makefile that lets me do things like make test and make release without having to remember what particular tools or additional steps this one project has.)

+1 for closing (or even removing the Makefile entirely).

Yep, done.

(IMHO in a Python project Makefiles are primarily a tool for maintainers. They make sense when the maintainer finds them useful. I find them useful: all of my personal Python projects have a Makefile that lets me do things like make test and make release without having to remember what particular tools or additional steps this one project has.)

While I understand where this is coming from, adding Make as a dependency to a project, even if only for maintenance, adds yet another hurdle to adoption for new people. They need to learn another tool that's not part of the standard programming environment. Python (and especially distutils and friends) is complicated enough without having people figure out Make on top of that...

I would actually agree if Makefiles were used as tiny wrappers on top of setup.py to remind people of the right targets to use. But looking at our own Makefile, it's not really the case: it's 200 lines of business logic, doing everything from a range of linting, unit tests, debian package build (!), homepage generation. It's a mess! And that's often what happens in those cases: the Makefile is there and then it's easy to expand it to do all sorts of random things and it grows and grows until it supplants setup.py for even stuff that it's supposed to be doing (like setup.py test).

This is arguably because packaging is too hard in Python. But that's being improved and we should follow that instead of hiding it under a layer of Makefile-ing.

So while I also do like Makefiles and use them often, I think that, in our case, we should totally get rid of it and do everything in setup.py, like good Python citizens. ;)

But until we go there, i think this can be closed as well. :)