It has a marvellous undocumented feature. Set the environment variable PIP_DOWNLOAD_CACHE to prevent re-downloading the same packages repeatedly when setting up environments on the same machine:
> set | grep PIP PIP_DOWNLOAD_CACHE=C:\Documents and Settings\jhartley\.pip_download_cache > pip install mock Downloading/unpacking mock Creating supposed download cache at C:\Documents and Settings\jhartley\.pip_download_cache Downloading mock-0.7.0b2.zip (242Kb): 242Kb downloaded Storing download in cache at c:\documents and settings\jhartley\.pip_download_cache\http%3a%2f%2fpypi.python.org%2fpackages%2fsource%2fm%2fmock%2fmock-0.7.0b2.zip [snip] Successfully installed mock > pip uninstall mock [snip] Successfully uninstalled mock > pip install mock Downloading/unpacking mock Using download cache from C:\Documents and Settings\jhartley\.pip_download_cache\http%3A%2F%2Fpypi.python.org%2Fpackages%2Fsource%2Fm%2Fmock%2Fmock-0.7.0b2.zip [snip] Successfully installed mock
(This text is copied from my unholy bastardised shell of choice at work, Windows CMD shell with Cygwin binaries foremost on the PATH.)
Using the download cache like this is substantially faster. Exactly what you need if you're continually setting up environments under various version of Python for testing or what-have-you.
The directory is created if it doesn't exist. Network access is still required when installing like this, presumably for the version checks.
Thanks to the irrepressible fuzzyman for bringing this to my attention.