[Discussion] How to update data?

[ckoerber]

Topic: How to update data

How shall this module allow users to update data? Possible scenarios are

1. Direct update of the central DB through registered users in the admin interface or related forms.
2. Storage of successive updates in file format applied to specific DBs after reviews (e.g., PRs)
3. ...

Challenges

1 Web-form updates / Central DB

Pro

• It is the most straightforward way of updating data and thus can be easily distributed among users

Con

• This requires that trusted users actively perform admin tasks and may be less transparent on default. E.g., who made a decision on which purpose.
• It also imposes a challenge for version control of the data. If a central DB changes without users informing users, this may have unexpected consequences.

Successive file-based updates / decentralized DBs

Pro

• File-based data updates committed to the repo make it more transparent when something was added and thus simplifies version control of data.
• If presented cleanly, updates allow proper peer reviews within the GH framework.

Con

• Updating data is more complicated. One needs to provide custom functionality, which allows users to provide continuous updates to the database. For example, check what my current status is and how it compares against the repo status.
• It puts more workload on admins to review/present updates

[ckoerber]

Idea for intermediate solution:

Providing yaml-like data migration files which can be utilized by get_or_create_from_parameters. This needs a template though.

[ckoerber]

The strops/schemes/scripts/create. Now create rough examples of how such scripts / yaml files might look like. With some more work, one may abstract this idea and write a general cli for new schemes.

[ckoerber]

Commit ef3b59c adds the script strops initdata which chains all init scripts in the right order.