SubformState should complain when the subform doesn't have #tree = TRUE

Well, thank you. You just saved me some time trying to figure out why one of mine didn't work :) After adding #tree I can finally see the values.

Cool!

Tagging as Novice, maybe someone will find time to fix this!

and partially unrelated, file_save_upload will not work with a subform. Apparently the $form_field_name expects the container name instead of the file field name.

Soham Sengupta → made their first commit to this issue’s fork.

I have added the fix, now the method createForSubform() throws an exception if the #tree property is not set, for non empty forms.
Moving the issue to Needs Review state.
Thank you.

Will need test coverage

Looks like we already have inadvertent test coverage! :D

All of SubformStateTest is failing because the form arrays aren't setting #tree. So those need to be fixed, and there should be one extra test added there which doesn't set #tree and uses expectException().

Fixed the unit tests.

I suspect most of the functional test failures are caused by the one form which I've fixed. Let's see!

Can issue summary be updated to the standard template, to mention the proposed solution is to throw an exception.

Then can mark.

Do wonder if this will break any contrib code and have to do the BC dance.

Without #tree, your subform is broken. I don't know how config field forms were managing not to be!

That's fair, then throwing exception makes sense.

If #tree is required then this needs to be documented, and I can not find anything about that. Can we have a followup for that?

Without #tree, your subform is broken. I don't know how config field forms were managing not to be!

But they are. So this potentially will trigger exceptions against currently working code. Because it did against core. I don't think we should do that. Perhaps we can start with a triggering a warning and then move to an exception in the next major version.