Example for squashmigration not working properly

[Sebastian Jekutsch]

Hi there,

sorry for the unspecific summary, but I expect to have a few issues here without knowing how to tell one from the other.

I expect squashmigrations with optimization turned on (default) to result in a squashed migration file so that running makemigration afterwards doesn't detect any migration needs. In my case though, makemigration detected for differences, i.e. some single migrations doesn't seem to be in the squashed migration file.

Squashing is important to me, because for some time I temporarily experimented with third-party libraries I don't want to include anymore. But installing/migrating in a new database still require them to be in the virtual environment, of course, if not squashed with optimization on.

See attachments:

• Zip of single migrations 0001 to 0116
• Resulting squashed migration file of "manage.py squashmigrations knowledgebase 0116"
• Resulting migration file of "manage.py makemigrations" -> 0117

Additionally, I found that two models seem to have errorneously disappeared completely in the squashed migration file as well as they have not been recognized by the makemigrations cross-check (which I expected to be empty anyway):

• Activity
• AbstractEntity - this may be ok since it's an abstract super-class, but note that AbstractCategory *is* included in the squashing

See attachment models.py

In case any information is missing please don't hesitate to contact me.

[Mariusz Felisiak]

Thanks for this report, it's a duplicate of #26223 (document.created_at, document.abstract, evidence.reference) and #32263 (actor.category).