Python3 RNG produces different results as compared to python2
In capa problems we save seed value and generate random state
using the seed. Python3 random generates a different state as
compared to python2. Thats why using random2 library to make
state genration consistent in both python versions.
* Update Financial Assistance logic
Use the zendesk proxy app instead of the unsupported zendesk library.
* Move to pre-fetching the group IDs.
Rather than making extra requests to zendesk to list all groups and find
a specific group ID. Just make a pre-filled list of group IDs for the
groups we care about. When a group name is passed in, it is checked
against this list and the ticket is created in the correct group so the
right people can respond to it.
* Convert to using ExitStack instead of nested.
nested is not available in Python3.
* Use ExitStack from contextlib2.
ExitStack has not been backported to the Python 2.7 standard library and
contextlib.nested is no longer available in Python 3.5. We need to be
able to use something in both 2 and 3 as we transition so we'll pull in
contextlib2 which has backports of the full python3 contextlib.
The old Notes Django app is the only caller of this no longer
supported library. I have verified that this old app is no
longer functional and is already slated to be removed as part
of DEPR-18.
So this commit simply removes the call to firebase-token-generator
and leaves removal of the Notes app to DEPR-18.
Pin to the minor version so that we can get patches via the update
automation. Since Django is pretty good at SemVer we can rely on these
to be pretty safe and it will speed up our security update process when
Django releases new security patches.
* Install drf-yasg
* Add drf-yasg settings and urls
* Pin drf to make drf-yasg work
* Adjust config-models version to be compatible
* Remove django-rest-swagger (the old way)
* Install drf-yasg
* Add drf-yasg settings and urls
* Pin drf to make drf-yasg work
* Adjust config-models version to be compatible
* Remove django-rest-swagger (the old way)
This has the effect of preventing Open edX installations from having
details which should be in configuration (because they're
edx.org-specific) end up in their applications.
We intend for subsequent versions to be reviewed manually before
finding their way into edx-platform proper; this prevents subsequent
releases from finding their way in without human intervention
