Regression from [#1038316] Allow for deletion of a single value of a multiple value field

Based on the steps what's covered here that was not in https://www.drupal.org/project/field_config_cardinality/issues/3205212 ✨ Restore Paragraphs' Collapse and Drag & Drop functionality Needs review

In template_preprocess_field_multiple_value_form() $item['_actions'] is checked for existence, and carried forward into cell 3 of 4 in the row cells/'data'. field_config_cardinality is out of step with this more recent change to core, which is what patch #2 addresses.

Note that this change was only introduced in Drupal 10.2. See https://www.drupal.org/project/drupal/issues/1038316 📌 Allow for deletion of a single value of a multiple value field Fixed For the 4.x branch of this module, that's not a problem as minimum version is already higher than 10.2. But for a backport to 3.x, compatibility with Drupal 10.1 and below - which did have this code for an actions column - would have to be considered.

I'm experiencing the symptom stefan.korn mentioned of tabledrag show/hide for row weights not working properly, on a Drupal 11.2 site. At least on this particular 11.2 site, patch #2 is not enough to restore the show/hide functionality. I'm currently looking into what else might have changed in template_preprocess_field_multiple_value_form() in recent releases, maybe there is something else to also take into account.

The further trouble with weights show/hide seems to be related to this change released in core 10.4 and 11.1. $order_class changed to be based on the table ID, which uses Html::getUniqueId() to assure uniqueness.

I see in theory why that would put field_config_cardinality out of step once again. And I think the reason it's such a universal problem to FCC-altered fields in practice is that getUniqueId() transforms underscores to dashes - e.g. $order_class = 'field-name-delta-order' - while the old approach of using the field name didn't transform anything and so $order_class = 'field_name-delta-order'

MR 20 is ready for review and should fix both regressions, and associated, for FCC 4.0.x.

Backport to 3.x would require putting the 'actions' fix inside conditionals to keep the old behavior on Drupal 10.1-

The MR does nothing to address stefan.korn's point about, does the module actually need to duplicate all the steps that it does from Core's theme function. Afraid I don't have time to dive into that myself.

I've faced the same issue, but I've come to a different solution before I've seen the MR.
Instead of trying to replicate the core behavior, my patch simply extracts the class name produced by core.

Like to try a solution that doesn't require checking core version if possible.

Like to try a solution that doesn't require checking core version if possible.

That's exactly what's done in #12.
Do you want me to create the MR?

Almost the entire issue queue seems to be bugs and probably because it lacks test coverage. I'm working on adding some basic coverage so we can wait on those are start new ones too

Seems to be failing 10.5 will have to figure that out then we can merge.

This is the last bug on the list and we can do another release

Automatically closed - issue fixed for 2 weeks with no activity.