- The Mirror was completed the total size of the mirror was around 150 GB.
- To Build the Dependency graph, conda-skeleton downloads the package, tries to run setup.py and see what package is missing. I definitely need to do something smarter. BTW there have been some attempts to build dependecy graph of PyPI, like this. But the author build dependecy using the
install_requires in setup.py but the problem is not that often not all dependencies are mentioned in install_requires, well I don't know if I can do any better. Before going with the dependency graph I'll create the download statistics.
- I was also able to use
conda-skeleton from my local PyPI mirror because I couldn't make xmlrpc requests to the mirror. I'll probably have to run the PyPI server on the mirror, there should be one available publically.