Special case dependencies need to be installed first. Historically, the two biggest offenders in terms of dependencies have been PyTables and Numpy. PyTables, until only recently could you install through pip reliably but still fails to install if you try to install it as a python package dependency (which is admittedly doesn't make much sense and I've found and reported the bug). Numpy simply will not install with older versions of pip and, as I've been told, historically didn't play well with python packaging until somewhat recently as well.
EricR86
commented
8 years ago
Original report (BitBucket issue) by Eric Roberts (Bitbucket: ericr86, GitHub: ericr86).
Special case dependencies need to be installed first. Historically, the two biggest offenders in terms of dependencies have been PyTables and Numpy. PyTables, until only recently could you install through pip reliably but still fails to install if you try to install it as a python package dependency (which is admittedly doesn't make much sense and I've found and reported the bug). Numpy simply will not install with older versions of pip and, as I've been told, historically didn't play well with python packaging until somewhat recently as well.
Specifically this issue is the main hangup at the moment: https://github.com/pydata/numexpr/issues/176
PyTables requires numexpr to be installed. Numexpr itself cannot be installed as a dependency and PyTables by extension either.
It might be possible to simply have numpy installed first before running
pip install segway