Memory usage

Record objects have a low memory footprint because they store attributes using slots rather than a per-instance dictionary. They use significantly less memory than an equivalent dictionary and slightly less memory (8 bytes to be precise), than an equivalent namedtuple. The example below was executed in Python 3.4:

>>> from reck import make_rectype
>>> from collections import namedtuple
>>> import sys
>>> Rec = recktype('Rec', ['a', 'b'])
>>> rec = Rec(a=1, b=2)
>>> NT = namedtuple('NT', ['a', 'b'])
>>> nt = NT(a=1, b=2)
>>> dct = dict(a=1, b=2)
>>> sys.getsizeof(rec)    # Number of bytes used by a record
56
>>> sys.getsizeof(nt)     # Number of bytes used by a namedtuple
64
>>> sys.getsizeof(dct)    # Number of bytes used by a dict
288

These memory savings are usually not significant unless you have a very large number of instances (e.g. hundreds of thousands), or are working on a low memory device.