Seeding scripts

[pbanaszkiewicz]

This fixes #2239 by providing an alternative to fixtures from Django.

Seeding scripts give more control to the developer. This way it was possible to implement a check before given entity gets created.

---

• To see the specific tasks where the Asana app for GitHub is being used, see below:
  • https://app.asana.com/0/0/1203911669015361

[elichad]

The deprecate_models method doesn't work for me if I add names to DEPRECATED_X variables - suggestions should fix it. Looks good otherwise.

Does the seeding get run automatically during the deployment process?

[elichad]

Suggested change

	if not model_class._default_manager.filter({lookup_field: model_id}).exists():
	if not model_class._default_manager.filter(**{lookup_field: model_id}).exists():

[pbanaszkiewicz]

Good catch! Thanks, fixing now.

[elichad]

Suggested change

	model_class._default_manager.filter({lookup_field: model_id}).delete()
	model_class._default_manager.filter(**{lookup_field: model_id}).delete()

[elichad]

If we wanted to update an existing entry (e.g. change the text of an email template), how would we do that with this setup?

[pbanaszkiewicz]

@elichad

Does the seeding get run automatically during the deployment process?

Yes, it's part of Dockerfile CMD. When container starts up, start.sh script is invoked. This script runs some commands before starting a webserver:

1. checks
2. migration
3. createcachetable
4. now seeding
5. superuser create (in another PR)
6. finally a webserver.

If we wanted to update an existing entry (e.g. change the text of an email template), how would we do that with this setup?

That's a great question. It should be agreed with product users that they accept potential overwrite of their changes. In this case, I assumed that no changes to existing email templates should be lost as a result of seeding.

I'm using a very similar setup to update entries during seeding, the change lies in these lines (seed_models()):

        if model_class._default_manager.filter(**{lookup_field: model_id}).exists():
            _info(f"{i} {class_name} <{model_id}> already exists, skipping.")
            continue

When existing object is found, the code makes an update of specific fields from model definition instead of skipping.