Map models to SQL
ZODB has been good so far for prototyping. It has helped write proper classes for the data. Things however are becoming more intrincated with ZODB with too many relations and complex queries. The data is also highly dynamic, much more than anticipated because of the filters.
There are also no arbitrary depth in hierarchies so mapping the current model to SQL is trivial and SQLAlchemy can handle relationships easily. The "exes" should be assembled directly from SQL queries and an agressive caching policy will have to be used to make it fast.
The current models have to be made independent of ZODB (trivial, just don't subclass from persistent) and a new package mapping the core models with SQLAlchemy mappers have to be written.
This app can be easily implemented with routes instead of traversal. The security levels will be at the experiment level and but I don't see a disadvantage on using traversal. The only disadvantage could be that it makes it harder for people not familiar with traversal.
Blueprint information
- Status:
- Not started
- Approver:
- None
- Priority:
- High
- Drafter:
- None
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- Approved
- Series goal:
- None
- Implementation:
- Unknown
- Milestone target:
- None
- Started by
- Completed by