[Data Objects] Save default values to versions

[BlackbitDevs]

With #5594 saving field's default value to database was introduced. The problem is that these default values do not get saved to versions currently. This is especially bad because when you publish this version, you will again have null in the respective fields.

Steps to reproduce:

1. Create class with input field abc, default value test
2. Create object of above class and save it
3. In the database tables now there is test in column abc
4. Go to versions tab, select jsut saved version -> there abc is empty
5. Restore the just saved version -> The value is now also null in the database column -> this is really bad because you can not restore the exact state the object had at this point (what is the main purpose of versioning beside tracking changes)

With this PR the default values are also saved to the database (and also to the PHP object being saved - which is important, if you have further operations with it after saving).

[BlackbitDevs]

Are the failing tests caused by this PR? I don't think so but am not sure

[brusch]

@BlackbitNeueMedien failing tests seem to be related, base branch runs through without any issues.

[weisswurstkanone]

@BlackbitNeueMedien I guess a unit test would definitely make sense here. wdyt 😉

[weisswurstkanone]

@BlackbitNeueMedien This needs to be done for fieldcollections, bricks & blocks as well.

[BlackbitDevs]

Added test, added support for localized fields, object bricks, field collections.

[BlackbitDevs]

@weisswurstkanone For the test I wanted to add an input field with a default value. How does this work? Obviously it is not enough to add it to the definition file in https://github.com/pimcore/pimcore/pull/10087/files#diff-07e12400f01e9028add55e187301705004f356fbc4a4c23ce29b2113a31d88e9R572

But CI complains about [Exception] Call to undefined method getInputWithDefault in class Pimcore\Model\DataObject\Unittest

[dvesh3]

@BlackbitNeueMedien you have to add a datachild with name inputWithDefault (and call setDefaultValue) in Unittest class setup here https://github.com/pimcore/pimcore/blob/10.x/tests/_support/Helper/Model.php#L549

[weisswurstkanone]

Thanks Divesh.

@BlackbitNeueMedien just for clarification: The classes are generated by code. The json files are just generated for your convenience.

[weisswurstkanone]

@BlackbitNeueMedien what is the status regarding the tests ? 🤔

[BlackbitDevs]

The status is that I currently do not know why the tests fail. If you have some time to support, this would be great.

[weisswurstkanone]

I am on it! Thx!

[weisswurstkanone]

Hello @BlackbitNeueMedien , regarding the test. I believe it should be

        $data = [
            'lblockadvancedRelations' => new BlockElement('lblockadvancedRelations', 'advancedManyToManyRelation', $reference),
        ];
        $source->setLtestblock([$data], 'de');
        $source->save();

        //link source on target
        $target = TestHelper::createEmptyObject();
        $target->setHref($source);
        $target->save(); //block data should retain here

        //update block element - manyToManyRelations
        $referenceNew = TestHelper::createEmptyObject();
        $referenceNewId = $referenceNew->getId();
        $referenceNew = new BlockElement('lblockadvancedRelations', 'advancedManyToManyRelation', $referenceNew);
        $source->getLtestblock('de')[0]['lblockadvancedRelations']->setData($referenceNew);
        $source->save();

so lblockadvancedRelations instead of blockmanyToManyRelations.

[brusch]

@BlackbitNeueMedien I'll close this PR since you're separating them anyway, right 😊 Looking forward to the 2 PRs.

[BlackbitDevs]

@brusch No. Please reopen it. I just remove the ValidationException things from this PR here and put those changes into another one.

[brusch]

@BlackbitNeueMedien sure, no problem 😊

[BlackbitDevs]

Ok, I reverted all changes related to the ValidationExceptions. This PR is now ready for review

[BlackbitDevs]

Sorry, @weisswurstkanone's fix got missing because I am too stupid to use git ;-)
Added it again.

[brusch]

@BlackbitNeueMedien I wouldn't introduce this change in a bug-fix release, because someone could rely on the current behavior. I think it would be ok to do this in a minor release, but before we gonna merge this PR we should discuss this as well.

In the meantime, could you please rebase this PR onto 10.x. Thanks a lot!

[BlackbitDevs]

Rebased to 10.x.
Actually it really is a bug because you cannot restore a certain state by publishing a version like I described in #10087 (comment) - but I do not have any problem that this gets released with the next minor release.

[CLAassistant]

All committers have signed the CLA.

[brusch]

@BlackbitNeueMedien one more minor thing, we should mention the behavior change in the upgrade notes. Thanks!

[BlackbitDevs]

@brusch I added documentation in the upgrade notes.

[weisswurstkanone]

thanks a lot!