Regression from #2521800: using machine name element for ListStringItem breaks with existing data

Added an MR, no idea if it's this easy but it might be.

Can we retain the code and just change the pattern on the #type machine name field?

That's a better idea, but what to? Spaces is the example given in the blog post, but presumably existing content could use more or less anything.

Actually thinking about #5 more, if we set replace_pattern to .* or whatever we come up with that will accept any existing data in the field, then the machine name widget essentially won't work - because it's just going to leave whatever input as is.

To make that work, we'd have to leave the regexp as is (or as is plus spaces maybe), but also remove validation for the machine name pattern when there's existing data in the field or something.

OK @lauriii's changes are implemething the last paragraph of #6 but prove it's actually easy to do whereas I thought it would be a pain. I can't see any drawbacks to that approach, so RTBC from me (and none of my code is in the MR now).

Pushed one commit to add an explanatory comment, leaving RTBC since it's just a comment addition.

alexpott → credited kevinquillen → .

This looks like a good fix.

Committed and pushed 7ca3f9e2e0 to 11.x and 35b420b9ee to 10.2.x. Thanks!

Neat solution 😍

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

🐛 Not able to add to List field with 0 as the first machine name. Needs work has a comment mentioning this issue.