On Thu, Jul 17, 2014 at 5:09 PM, Frank Millman <fr...@chagford.com> wrote: > He quotes some stats from PyPi, which shows number of downloads over a > period, broken down by version. Over a recent period, Python2 downloads > exceed Python3 downloads by a factor of 10:1 (subject to my memory ...)
These kinds of stats are always flawed. In a lot of cases, they're skewed heavily by defaults, and in other cases, skewed even more heavily by long dependency trees - so, for instance, a single Django installation might involve fetching large numbers of packages from PyPI, even though no new code has been written at all. Might add quite a bit to the download stats for one version or the other. And what about upgrades? Stable installations are still likely to want to get the latest, which means downloading from PyPI, ergo it's another hit. ChrisA -- https://mail.python.org/mailman/listinfo/python-list
