Closed marcrecurve closed 8 years ago
Numpy installation can be a bit sticky. Often we find ourselves using Anaconda.
It probably makes sense to standardize the dev setup, and Vagrant is probably the way to do that, in preference to Docker. Although it probably makes sense to consider setting up a base-level environment eventually, because there is definitely some overlap, and often we develop across the whole stack, let's stick with one vagrant file per repo at first to keep things simple.
Oh - can I have you pull request into develop instead?
Aiming to do this in a separate repo, one Vagrantfile for the whole stack.
More notes later. For now...
Vanilla install of dev env doesn't seem to work. Numpy compilation fails for some reason. Also there are some missing steps in the directions (e.g. assumes virtualenvwrapper is installed)
In past projects, have worked to encapsulating a full dev environment in Vagrant. The setup below allows you to do a
vagrant up
thenvagrant ssh
, and immediately run the tests. (Well, sort of...I need to clean the script up a little, but it's close.)The benefit beyond a single command for initial set up is that it unifies dev boxes across the org. No more asking "what version of python do you have installed?", since it's scripted in the provision of the Vagrant box.
Possible to do this with Docker instead of Vagrant. Also possible to use Docker as a driver for Vagrant.
Also seems worth considering where to create Vagrantfiles. One in this repo? One for the base level dev environment that you can install multiple OEE components into? Others?