[Bug]: PersistRelations only works on BelongsTo(Many) relations which are nullable

[kerwin-wildsea]

Is the bug applicable and reproducable to the latest version of the package and hasn't it been reported before?

• Yes, it's still reproducable

What version of Laravel Excel are you using?

3.1.63

What version of Laravel are you using?

11.43.2

What version of PHP are you using?

8.4

Describe your issue

Due to the ->save() call in src/Imports/Persistence/CascadePersistManager.php:44 which happens before the relations are persisted it is not possible to use the PersistRelations interface for relations which are required (NOT NULL).

How can the issue be reproduced?

1. Create two models with a BelongsTo relation between them
2. Make the relation ->foreignId(RelatedModel::class)->constrained() (NOT NULL)
3. Create a ToModel import including $model->setRelation('related', new RelatedModel);
4. Call import on the importer

What should be the expected behaviour?

I'd expect to either note in the docs that this is a limitation of the interface (a massive one) or that it'd not save the model before persisting the relations.

[patrickbrouwers]

If you have a look at the test https://github.com/SpartnerNL/Laravel-Excel/blob/3.1/tests/Concerns/ToModelTest.php#L169 , I'm doing BelongsTo differently than your example, have a look if doing it that way makes any difference. (Perhaps the correct example is missing in the docs then - if so, would appreciate a PR to the docs). If not please provide a failing unit test to the ToModelTest.

[kerwin-wildsea]

@patrickbrouwers Thanks for the response. I tried the method using associate instead of setRelation but it has the same result.

Due to the save() function being called before any relations are persisted, any NOT_NULL FK columns will cause the DB to throw a: Integrity constraint violation: 1048 Column '<col>' cannot be null

In the test, it works because the relation is nullable.

I could provide a test, but it'd require a new model + NOT NULL relation in the tests which would affect other tests as well (if the NOT NULL relation is added on an existing model). How would you approach this? Add two new models with different relation?

[stale]

This bug report has been automatically closed because it has not had recent activity. If this is still an active bug, please comment to reopen. Thank you for your contributions.

[patrickbrouwers]

Add two new models with different relation?

Yes I think that's the best idea

[KSneijders]

@patrickbrouwers

Yes I think that's the best idea

Thanks for the reply, I've created a PR (#4304) including a new model with a mandatory relation.
You can see in the CI/CD results that the tests fails due to the relation being null even though it is being set using setRelation.

When changing the migration to include ->nullable() the tests succeeds again.

Hopefully this is enough to show the bug I'm reporting in this issue

[KSneijders]

@patrickbrouwers Hey, friendly reminder to check out the PR (#4304) to see the new relation failing in action.

[stale]

This bug report has been automatically closed because it has not had recent activity. If this is still an active bug, please comment to reopen. Thank you for your contributions.