Created a CONTRIBUTING.rst which GitHub will pick up and included a
link to the How To Contribute wiki page there. Also added the same link
to the README in the "How To Contribute" section.
vagrant:
- check actual IP;
- only init machine first time, just as native installations;
- add .bash_profile to permissions fixup;
- cleanup excessive newlines in making .bash_profile;
README:
- include admin & debugging use;
- info about re-initializing VM & manually running create-dev-env.sh;
- add django admin & debug_toolbar setting info
Conflicts:
AUTHORS
scripts/vagrant-provisioning.sh
Instead, we use XModule field default values when creating an empty
XModule. Driven by this use case, we also allow for XModules to be
created in memory without being persisted to the database at all. This
necessitates a change to the Modulestore api, replacing clone_item with
create_draft and save_xmodule.
Cf README, this adds support to install and setup a dev instance through
Vagrant. The Vagrantfile at the root handles the vagrant commands like
`vagrant up`.
Provisioning is done by the shell script `vagrant-provisionning.sh`,
which is a light wrapper around the `create-dev-setup.sh` which does
most of the job using the standard process, but limiting possible
side-effect by ensuring the environment on which the install is
performed is clean.
Based on an Ubuntu precise 12.04 LTS.
When `pip install -r requirements/edx/pre.txt` is omitted before running
`pip install -r requirements/edx/base.txt`, it would cause API versions
mistmatch when `make cms` is run. `make cms` would automatically run pip
with pre.txt and install a different version of numpy, *after* base.txt
and post.txt have been run through pip.
