Hidden weight column becomes visible after upgrading from 3.0.x to 4.0.x

What version of Drupal core? That might be helpful to know. I believe this change was added to support newer versions of core.

MR using solution from @abramm to hopefully provide a D10/D11 compatible solution:
https://www.drupal.org/project/field_config_cardinality/issues/3454609#c...

I tested this patch with Drupal 10.5.4 and 10.5.6 and it works with both.

For the record, this patch ALSO fixes the issue on the same where the weight handles didn't appear, which makes sense looking at the issue and tracking down how tabledrag.js works.

For testing, we were able to reproduce this by testing with fields that were paragraphs; for these, the table name is not the same as the field name that core uses.

Thanks so much everyone!

I think this patch is much more critical than what's been described to date...without it, tabledrag was completely broken for me, whether using Claro or Gin (which has tabledrag.js overrides). My stack is core 10.6.3, paragraphs_browser 1.3 (using the "experimental" widget...though "classic" yielded the same error).

Pre-patch, I was getting this in the console:

tabledrag.js?v=10.6.3:525 Uncaught TypeError: Cannot read properties of undefined (reading 'matches')
    at tabledrag.js?v=10.6.3:525:22
    at Array.map (<anonymous>)
    at Drupal.tableDrag.rowSettings (tabledrag.js?v=10.6.3:522:8)
    at Drupal.tableDrag.updateField (tabledrag.js?v=10.6.3:1083:28)
    at tabledrag.js?v=10.6.3:1070:12
    at Array.forEach (<anonymous>)
    at Drupal.tableDrag.updateFields (tabledrag.js?v=10.6.3:1067:43)
    at Drupal.tableDrag.dropRow (tabledrag.js?v=10.6.3:912:14)
    at HTMLDocument.<anonymous> (tabledrag.js?v=10.6.3:311:57)
    at HTMLDocument.dispatch (jquery.min.js?v=3.7.1:2:40035)

I eventually wound up at this issue because I realized that the "field_name-delta-order" class was not being added to the last table cell (@see template_preprocess_field_multiple_value_form(), Drupal.tableDrag.prototype.rowSettings) -- the class gets removed after tabledrag.js processing has run, but without its presence initially, the drag target can't be found.

Heads-up: if patch application fails with “malformed patch”, check the line endings on the downloaded .patch file. If it was saved or edited on Windows (or a tool rewrote it), it may have CRLF (\r\n) instead of Unix LF. BSD/macOS patch is picky about that and can report a malformed hunk even when the diff content is fine.

Trying to improve the title

Neither the patch file in #2 nor the MR fix the issue for me.

Also, #11 set this issue to RTBC but described issues applying a .patch file; if the MR had been used it would've been a .diff file. The patch and the MR seem to take different approaches to resolving this bug, but to me it doesn't seem fixed anyway. I'll see if I can replicate on a fresh sandbox environment.

Nevermind; PEBKAC -- I had an extra trailing slash in my composer.patches.json file.

I've confirmed both the patch and MR resolve the issue, though the MR seems to be more thorough.