XBlockAsides
...
Uncertainties
Dependencies
Django upgrade to 1.6 (or maybe 1.7, depending on what ops says about in-progress password reset requests)
Uncertainties
How many users have in-flight password reset requests? Do we need to wait for them to clear between going from 1.6 to 1.7?
- We have our own password reset logic that wraps the django logic
- Do we need our subsidiary libraries (submissions, openassessment, xblock, maybe others) to support multiple versions of django?
...
There are many ways to unexpectedly shoot ourselves in the foot with bad call patterns into modulestore (or bad access patterns when working with XBlocks). We should do some discovery on how to make it easier to fall into a pit of success when using XBlocks, performance-wise.
Discovery: XBlock Migrations
We have a number of XBlocks that have (for instance) fields with the wrong scope, or many fields for the same data to handle old version of the data in storage. Having a framework for isolating the code that needs knowledge of old formats would make it significantly easier for other teams to make changes to XBlocks without having to worry about backwards compatibility in future code.
The Video and Capa XModules are particularly bad instances of this.