Changing a field name in the way that we did (updating the Python variable name and switching it to point to the old DB column) confuses Django's migration autodetector. No DB changes are actually necessary, but it thinks that a field has been removed and a new one added. This means that Django will ask users to generate the migration it thinks is necessary, which ended up with us dropping data. The fix is to run the same migration on Django's internal model of the DB state, but not actually touch the DB.
Name |
Last commit
|
Last update |
---|---|---|
.. | ||
0001_initial.py | Loading commit data... | |
0002_coursemode_expiration_datetime_is_explicit.py | Loading commit data... | |
0003_auto_20151113_1443.py | Loading commit data... | |
0004_auto_20151113_1457.py | Loading commit data... | |
0005_auto_20151217_0958.py | Loading commit data... | |
__init__.py | Loading commit data... |