Re: Why do we need Python sub-modules?

[Adam <adam%NetBSD.org@localhost>]

>> I propose we enable back at least some of the modules, like those
>> that do not require too many dependencies (I would keep 'tkinter' in
>> py-Tk, for example). That would keep Python as standard as possible.
> 
> It's been a while since you asked this, but I am all for reducing
> complexity. If things got subsumed into standard modules for both
> python2 and 3 and are stable there, I would like not having to think
> about separate packages. And it reduces hackery to disable them in the
> python package, perhaps. The complication I see is that things need to
> be symmetric between python versions, as long as we support 2.x and 3.x
> versions at the same time. At least I have my build script try to build
> both 2.7 and 3.6 versions of all modules.

In couple of days I would like to change all Pythons to include missing modules: curses, curses_panel, elementtree, sqlite3, pyexpact, readline, spwd, Again, Python modules do not list them as dependencies (why would they?), but assume they exist, so it is hard to properly package Python with the current setup.

I will remove module packages (py-readline, py-curses, ...) and bump revision for packages that depend on them.

I am also going to provide fixes for building with curses (based on latest Python patches) and finding libraries with ctypes.util.find_library.

If someone has a strong reason for not doing these changes, please, speak now. :)

Kind regards,
Adam