Skip joined entity creation for empty relation (#10889)

[noemi-salaun]

(The following is translated from french with Google Translate, I'm sorry :/ )

In case the root entity has its primary key as a relationship and the joined entity has a simple primary key, the RSM orders the ID of the joined entity before having the ID which defined the root entity.

The ObjectHydrator therefore first tries to hydrate the joined entity. We then fall into a special case where the parent (the root entity) has not yet been created and the hydrator tries to start by creating the child, except that in our case, the child is supposed to be null.

I therefore added the isset($nonemptyComponents[$dqlAlias]) check which we find further in the code and which is made to skip the creation of the joined entity when the parent is already hydrated. I think it makes sense to do the same check in the case where the hydrator starts with the joined entity.

[mpdude]

What does it mean when the GH10889Hand class has an @Id column that is a one to one association and a generated value at the same time?

[noemi-salaun]

ah you're right it's an oversight. During my tests I wanted to see how the ObjectHydrator behaved with a simple ID rather than a relationship, and I left an annotation. It doesn't change anything about the problem here but I'm going to remove it ASAP

[noemi-salaun]

@mpdude It's corrected. I checked, I still have the same problem without the GeneratedValue annotation, and the problem is still fixed with my changes.

[noemi-salaun]

I don't think the failed psalm static analysis is related to my changes

[noemi-salaun]

@mpdude
Any info on when these changes might be merge? I am still locked to v2.15 in my project because of this

[Prokyonn]

We've tested the changes from this PR in Sulu, everything seems to be running without encountering any issues.

However, until this is merged, we also have to lock the doctrine version in our customer projects to v2.15. Given this situation, we would greatly appreciate it if you could prioritise merging this PR.

[derrabus]

if you could prioritise merging this PR.

The PR cannot be merged due to conflicts.

[winiarekk]

@noemi-salaun will you be able to resolve these conflicts?

[noemi-salaun]

@derrabus I just rebased my work on the 2.19.x branch, it should be good now

[noemi-salaun]

@mpdude @derrabus
Is there still something blocking this PR?

[noemi-salaun]

@greg0ire I changed the test with entities with less meaning

[greg0ire]

Thanks @noemi-salaun !