datetime is a wonderful module. Perhaps the only inconvenient aspect of
using it is dealing with month calculations and comparisons. This patch
adds a simple class, monthdelta, which represents date offsets in terms
of months. It supports basic integer-like arithmetic, and also it may
be added to dates and datetimes. It deals sensibly with leap-year and
month-length issues. Also provided is a function, monthmod (named by
imperfect analogy to divmod), which allows round-tripping: that is,
taking 2 dates and returning a (monthdelta, timedelta) tuple that
represents the interim between the dates.

Note: I have named the class "monthdelta", but in light of recent
python-dev discussions I should probably rename this to "MonthDelta".

This is my first try at a patch. All functionality, tests, and
documentation are included, but I won't be surprised if I need to make
some changes! Please let me know.

Rietveld link: http://codereview.appspot.com/25079

This prototype python implementation passes the same tests that the C
implementation in the patch does (modulo import differences). I'll
probably backport this for 2.x versions.

With the rejection of bpo-5530, it seems best for the name of this
class to remain lowercase. Mixing casing schemes within the same module
would be perverse.

The backport through python 2.3 is up at PyPI:
http://pypi.python.org/pypi?name=MonthDelta&:action=display

Download source and install, or just easy_install. Documentation is
also linked at PyPI.

MonthDelta has been published on PyPI and there was no activity on this issue since then. I conclude that there is little support for including this in stdlib. I am marking this as "rejected" and setting the status to pending. I will close this issue in a few weeks unless I hear objections.

Jess,

I see that your package is healthy on PyPI with over 700 downloads for the latest version. Note that some additional operations have been added recently to timedelta object and more are in the works. You may want to consider adding them to the next version of your package. (See bpo-1289118 and bpo-2706.)

Yeah, the masses have spoken, and their silence is deafening. I was going to downplay the "health" of the package based on the number of downloads, but then I looked at the "futures" page. b^) Seriously, getting this in would require a core developer wanting it in, and at this point it seems none do.

On the subject of float operations, while there may be some question about the integerness or floatness of timedelta, monthdelta is definitely an integer. As monthdelta already supports multiplication by integers and floor division by integers or monthdeltas, I don't see anything missing. To use monthdeltas with floats is to miss the point entirely.

Rejecting this for stdlib. A 3rd party package is available at http://pypi.python.org/pypi?name=MonthDelta&:action=display .

For future reference, python-dateutil seems like the maintained way to deal with "human" dates outside of the stdlib.

http://pypi.python.org/pypi/python-dateutil