2

I have a Flask app with a dependency on PyTables that needs to go onto a production server. Unfortunately, the script that must be used to install all dependencies only works if everything can be pulled in by pip install -r requirements.txt. In other words, running scripts / using setup.py is not possible.

PyTables requires numpy at install-time to build (as well as numexpr and cython), so just putting the dependencies sequentially in requirements.txt yields ERROR:: You need numpy 1.4.1 or greater to run PyTables!.

My hope is that there is some way to just pre-install the dependencies into their own repo and point to that in requirements.txt. This seems hack-ish, and I'm very open to alternative suggestions. Thanks!

brice
  • 21
  • 1
  • 1

1 Answers1

1

I believe the best course of action is to create multiple requirements files. If you include dependencies for other packages in a requirements file you install first, you should be able to get past your error.

See here for directory structure: https://stackoverflow.com/a/20720019/3050804

Community
  • 1
  • 1
Pythagoras
  • 46
  • 2
  • As a test, I made a requirements directory with table_requirements.txt in it (containing numpy, cython, etc.), and added `-r requirements/table_requirements.txt` to the top of requirements.txt (in the parent directory). Is that what you meant? I am still getting the same error, unfortunately. – brice Oct 21 '14 at 23:30