Models: OCD or our own from scratch? about am-django-project

[gordonje]

Maybe it's overconfidence, but I feel comfortable mapping out these domains, including campaign finance and generic people. And having control of these models actually feels more like a timesaver in terms of making changes faster without having to consider the broader scope (e.g., does this new field make sense in every legislative body?) Also, we would likely end up dealing with the cruft and terminology translation you mention in am's own views and admins.

We should probably investigate in more detail what exactly integration/interoperability would look like. If this is more of a secondary or long-term maintainability concern, maybe the middle ground solution would be something like an app in the am-django project for loading from/into opencivicdata models. This would allow us to keep control over mapping out the domains relevant to Access Missouri while creating a space for dealing with integration if/when those issues actually arise.

Are OCD's django models so close to being exactly what the Access Missouri needs that we don't mind maintaining what little grafting logic is required as part of AM's core (e.g., in the views and admins)?

If not, then is integration/interoperability a primary concern? (And I feel like it is if we can't name the specific services with which we're integrating).

If not, then I say we should definitely be writing our own models. We could start by pointing django at the current access mo postgres db and doing inspectdb and making some edits inspired by OCD's specs.

This feels like a 180 for me. Sorry bout that.

[NathanLawrence]

I'm coming around to this as well. In terms of using inspectdb, we have some problems that would prevent us from doing that, but the current structure is such a mess that I'm not sure I'd want to take that route anyway. I have posted this for you, though -- it's the PeeWee version of all our current models based on my inspection in a gist.