Chris Broz

We've documented all planned fixes. Only lingering issue seems to be varchar changes on exports, for which there is an available script

**Related Issue**: It is also the case that many of our `IntervalList` inserts use `replace=True`. This makes sense to update a conceptually linked, but non-fk-ref'd item, but relies on each...

The team discussed a partial solves that would involve respective pipelines checking related `interval_list_name`s for overlapping times to reduce redundancy

To fully address this issue in the distant future, it would be helpful to add an `IntervalListObject` class, separate from the table itself. This class would be responsible for validating...

Under the status quo, inserts to `IntervalList` use `replace=True`. This assumes the pk being used is unique to the inserting table. Because naming conventions are decentralized, it is unlikely, but...

I think (1) the selective dump of mysql is readily achievable. It might involve generating a list of the tables associated with paper data, and then using datajoint to generate...

Thinking through some ideas here with the help of diagrams ... ## Strategy In an effort to monitor an analysis and generate a corresponding export, I found I would identify...

We discussed recording an export based on `paper_id` and `analysis_id`. This will allow users to revise an export based on an individual notebook/figure rather than starting over. It seems likely,...

My current solution is to hook into `dj.restrict` with the mixin. I made some progress collecting and cascading restrictions to other tables, but I ran into the following issue: multiple...

Is used data always fetched? I imagined that there might be operations without fetching, but, no I can try to hook into fetch