[Sean1708]

My main issue with it is that if someone figures out an even better way to do dictionaries but it doesn't preserve insertion order, then that technique can't be used. But I certainly wouldn't call it a stupid decision, the problem is that once a behaviour is observable then people absolutely will start relying on it.

[mark-r]

All they have to do is give it a different name, such as unordered_dict.

[jmilloy]

I might have preferred odict be added and dict left alone, so that you didn't have to import OrderedDict when really you just want the new dict. If dict and odict mapped to the same thing for a while, fine, but dict could diverge again if desired.

[masklinn]

The problem is that when you behave a certain predictable and deterministic way users will leverage that even if it's not specified.

That is actually why the Python maintainers decided to make this behaviour official: the ordering was the consequence of changes in implementation details, but over 3.6's lifecycle they feared it would cause compatibility issues as users would start relying on the ordering properties of CPython and that would be an issue for alternate implementations (though pypy had switched to the same implementation even earlier so was also insertion-ordered even when running in Python 2.7 mode).

Providing stronger guarantees was considered useful and unlikely to be severely detrimental in the long run, so the project decided to make it part of the language spec.

[thedirt0115]

"users will leverage that even if it's not specified" -- reminds me of the classic xkcd, "Workflow": https://xkcd.com/1172

[mark-r]

An xkcd in the thousands is now considered "classic"? Man I'm getting old.

[compiler-guy]

Which means no one gets it for free when it is added to the language.

[jmilloy]

You could add an ordered dict by a different name and also have dict be the more efficient ordered version so that people get the efficiency gain for free but don't have to worry and forward compatibility.

[harikb]

I think this is a key point. Since they advertised this, it became a commitment for all future versions. Sure they can break again in 4.0

[jrockway]

Maintaining insertion order is strictly a product of adding an additional list to whatever more efficient map you implement. Add something to a dict, push it onto the end of the list. When iterating over the keys, use the list instead of the dict structure.

It does come at a cost, but I think Python aims for ease of use over runtime speed and memory efficiency, so it seems perfect for them.