It's still the case as of v5.30 that the incomplete UTF8 support is hard coded into the core code.

The solution for now seems to be to re-run the System.utf8conversion routine after any time you create a field/fieldset.

Incomplete UTF8 support is hard coded into the core code before 5.33 (thanks @michaelmcandrew!).

The solution seems to be to re-run the System.utf8conversion routine

• Pre 5.33: do this after any time you create a field/fieldset (including anytime you install an extension that creates a custom field/fielset)
• 5.33+ do this once, then the only time it should be needed again is if an extension hard-codes the old UTF8 in place.

Note that you can use the uh-oh extension to help discover whether your db uses the old UTF8 stuff.