Refs #27236 -- Clarified that "historical migrations" means "already applied".

[felixxm]

No description provided.

[carltongibson]

Yes, I think it clarifies the meaning somewhat. 👍

@collinanderson do you think this assists some?

[collinanderson]

Hmm... I'm not sure I like defining "historical migrations" as "already applied". What if I need to take an old database backup and apply migrations to it? Now those "historical migrations" aren't applied. Can I apply these historical migrations to an old database backup after 5.1? It's unclear to me what the behavior would be after after 5.1, if there's any change at all. (Does it work the same way? Is it now a no-op?)

[collinanderson]

Isn't the main point that deprecated migrations operations are never auto-generated by Django any more, and generally shouldn't be used in new migrations, but the migration operations won't actually be removed?

(If they are being removed, then it should say what replacement operation I should be using instead so I can go into my historic migrations and edit them to remain compatible.)

[felixxm]

What if I need to take an old database backup and apply migrations to it? Now those "historical migrations" aren't applied.

I'd argue 😉 history is something that has already happened 🏰. When migrations are not applied to an old database backup then they are not historical in this db.

Can I apply these historical migrations to an old database backup after 5.1? It's unclear to me what the behavior would be after after 5.1, if there's any change at all. (Does it work the same way? Is it now a no-op?)

No you cannot, when deprecation ends the current warning will be changed to an error (it's exactly the same behavior that we used in the past when deprecating model fields). That's why in the subsequent PRs I would like add descriptive release notes, implement reducing AlterIndexTogether with RenameIndex, and add a hint to the migrations.W001 that refers to extensive notes.

[felixxm]

or .... https://code.djangoproject.com/ticket/27236#comment:36

[felixxm]

Closing in favor of #15861.