Hi Simon,
Nope, you're not missing anything. I agree with reverting the fix in #25858
and going with approach #2 as outlined in your initial post.
Sorry that wasn't clear - I meant to address my response to the thread as a
whole, in light of that fix being found to cause another regression.
Chee
Hi Alex, thanks for chiming in.
Just to make sure I understand your second proposed option; I'm having a
hard
time understanding how it's different from simply reverting the fix for
#25858[1].
Are you suggesting that the abstract model's foreign key be resolved or the
copy.copy()'ied field atta
