python 3.15.0, pandas 3.1.0

[1. caching a function that takes a config]
dict       -> TypeError: unhashable type: 'dict'
frozendict -> 3 calls, function body ran 1 time(s); CacheInfo(hits=2, misses=1, maxsize=None, currsize=1)

[2. equality and hashing]
a == b: True | hash equal: True | a == dict(x=1, y=2): True
list(a): ['x', 'y'] list(b): ['y', 'x'] (insertion order kept)

[3. MappingProxyType vs frozendict]
proxy sees: EUR | frozendict sees: USD
proxy assignment -> TypeError: 'mappingproxy' object does not support item assignment
frozendict assignment -> TypeError: 'frozendict' object does not support item assignment
hash(proxy) -> TypeError: unhashable type: 'dict'
hash(frozendict) works: True

[4. default arguments]
dict default: [1, 2, 3]
frozendict default: [{'source': 'api', 'id': 0}, {'source': 'api', 'id': 1}]

[5. shallow immutability]
after append: frozendict({'features': ['age', 'income', 'zip_code']})
hash(cfg) -> TypeError: unhashable type: 'list'
with a tuple value, hash works: True

[6. updates create new objects]
base: frozendict({'lr': 0.1, 'depth': 6}) | tuned: frozendict({'lr': 0.05, 'depth': 6}) | type: frozendict
after |= : base: frozendict({'lr': 0.1, 'depth': 8}) | alias: frozendict({'lr': 0.1, 'depth': 6})
copy.copy returns the same object: True | .copy(): True
update() -> AttributeError: 'frozendict' object has no attribute 'update'

[7. working with other code]
isinstance dict: False | isinstance Mapping: True
json.dumps: {"name": "ada", "score": 0.93}
json.dumps(dict(fd)): {"name": "ada", "score": 0.93}
pickle round trip equal: True | type: frozendict
f(**fd): ada=0.93

[8. pandas: works, fails, or silently wrong]
  pd.Series(fd)        same
    dict:       {'a': 1, 'b': 2}
    frozendict: {'a': 1, 'b': 2}
  Series.map(fd)       same
    dict:       ['KR-11', 'KR-26', nan]
    frozendict: ['KR-11', 'KR-26', nan]
  rename(columns=fd)   same
    dict:       ['region', 'sales']
    frozendict: ['region', 'sales']
  astype(fd)           same
    dict:       'float32'
    frozendict: 'float32'
  fillna(fd)           DIFFERENT
    dict:       [3.0, 0.0, 5.0]
    frozendict: ValueError: Invalid fill value: expected scalar, dict, Series or DataFrame; got frozendict
  groupby.agg(fd)      DIFFERENT
    dict:       {'Busan': 0.0, 'Seoul': 3.0}
    frozendict: TypeError: 'frozendict' object does not support item assignment
  pd.DataFrame(fd)     DIFFERENT
    dict:       {'a': [1, 2], 'b': [3, 4]}
    frozendict: {0: ['a', 'b']}

[9. cost (timeit, best of 5, Apple M5)]
  build from a 20-key dict     dict   51.0 ns   frozendict   49.3 ns
  lookup d['k7']               dict    9.9 ns   frozendict    9.9 ns
  hash of a new 20-key frozendict  ~  73.0 ns (build+hash 122.9 minus build 49.9)
  hash of the same object again      13.8 ns
  hash again,   1,000 keys           13.6 ns
  hash again, 100,000 keys           13.5 ns
  sys.getsizeof: dict 464 bytes, frozendict 472 bytes