Squash migrations about dsmr-reader HOT 6 CLOSED

I will only squash dsmr_stats, reducing 16 migrations (files) to just a single one.

It also renders all runPython migrations obsolete, as Django will run the old migrations for checkouts halfway migrated and the squashed ones for new checkouts.

from dsmr-reader.

I will also drop any prints in legacy migration, as they are deployed everywhere by now. RunPython-migrations will be dropped completely in the squashed version, because they only fix existing data.

from dsmr-reader.

• ./manage.py squashmigrations dsmr_consumption 0004_recalculate_gas_consumption
• ./manage.py squashmigrations dsmr_datalogger 0005_optional_gas_readings
• ./manage.py squashmigrations dsmr_stats 0016_drop_stats_settings

Added a runpython for database init of MeterStatistics.objects.create().

from dsmr-reader.

I think I'm running into a migration bug in Django, but I'm not sure. It seems index operations for MySQL and PostgreSQL do not respect the operation order. When a squashed migration ends up having this:

• Create model
• Add field with index
• Drop model

Then the creation of the index is delayed after dropping the model, which is denied by the database. Only SQLite seems unaffected.

from dsmr-reader.

Seems related to:

• https://code.djangoproject.com/ticket/25577
• https://code.djangoproject.com/ticket/25530

I think the reason behind is that squashmigrations doesn't remove the model from the migrations. It shouldn't exist at all.

Fow now I will fix it manually by dropping db_index=True attribute of the model created and dropped later.

from dsmr-reader.

Merged & deployed.

from dsmr-reader.