Choosing Between OrderedDict and dict

00:00 Choosing Between OrderedDict and dict. For years, Python dictionaries were unordered data structures. Python developers were used to this fact, and they relied on lists or other sequences when they needed to keep their data in order. With time, developers found a need for a new type of dictionary, one that would keep the items ordered.

00:22 Back in 2008, PEP 372 introduced the idea of adding a new dictionary class to collections. Its main goal was to remember the order of items as defined by the order in which the keys were inserted. That was the origin of OrderedDict.

00:36 Core Python developers wanted to fill in the gap and provide a dictionary that could preserve the order of inserted keys. That, in turn, allowed for a more straightforward implementation of specific algorithms that rely on this property.

00:53 OrderedDict was added to the standard library in Python 3.1. Its API is essentially the same as dict. However, OrderedDict iterates over keys and values in the same order that the keys were inserted. If a new entry overwrites an existing entry, then the order of items is left unchanged.

01:11 If an entry is deleted and reinserted, then it will be moved to the end of the dictionary.

01:17 Python 3.6 introduced a new implementation of dictionaries. This represents a big win in term of memory usage and iteration efficiency.

01:27 Additionally, the new implementation provides a new and somewhat unexpected feature: dictionary objects now keep their items in the same order they were introduced.

01:36 Initially, this feature was considered an implementation detail, and the documentation advised against relying on it. In Python 3.7, the items-ordered feature of dict objects was declared an official part of the Python language specification. So, from that point on, developers could rely on dict if they needed a dictionary that keeps its items ordered. At this point, a question arises: Is OrderedDict still needed after this new implementation?

02:06 The answer depends on your specific use case and also how explicit you want to be in your code. Some features of OrderedDict still make it valuable and different from a regular dict. First, intent signaling.

02:20 If you use OrderedDict over dict, then your code makes it clear that the order of items in the dictionary is important. You are clearly communicating that your code needs or relies on the order of items in the underlying dictionary. Control over the order of items.

02:36 If you need to rearrange or reorder the items in a dictionary, then you can use .move_to_end() and the enhanced variation of .popitem().

02:45 Equality test behavior. If your code compares dictionaries for equality, and the order of items is important in that comparison, then OrderedDict is the right choice.

02:55 There’s at least one more reason to continue using OrderedDict in your code: backwards compatibility. Relying on regular dictionaries to preserve the order of items will break your code in environments that run versions of Python older than 3.6.

03:10 It’s difficult to say if dict will ever fully replace OrderedDict. At the moment, OrderedDict still offers interesting and valuable features that you might want to consider when selecting a tool for a given job.

03:23 In the next section of the course, you’ll get started by using OrderedDict.