> I don't have any criticism if the performance of this new dict stays the same as earlier and that's a huge if.
1. it's been there since 2016, the new dict was released as part of Python 3.6 (though the ordering was only an implementation detail until 3.7)
2. the new dict is much more compact (20~25%) and should have much faster iteration speed, those were actually the original goals, ordered iteration was a side-effect
3. the new dict should not benchmark significantly slower across the board, it wouldn't have been merged if it were a regression
1. it's been there since 2016, the new dict was released as part of Python 3.6 (though the ordering was only an implementation detail until 3.7)
2. the new dict is much more compact (20~25%) and should have much faster iteration speed, those were actually the original goals, ordered iteration was a side-effect
3. the new dict should not benchmark significantly slower across the board, it wouldn't have been merged if it were a regression