Objects created by a gettext_lazy/ugettext_lazy call occupies almost 12 KiB of memory.
This makes Django pretty unusable in an i18n environment with limited memory.
Here's a proof:
>>> import guppy
>>> heapy = guppy.hpy()
>>> heapy.heap()
Partition of a set of 68757 objects. Total size = 4651296 bytes.
Index Count % Size % Cumulative % Kind (class / dict of class)
0 25728 37 1694752 36 1694752 36 str
1 16723 24 615380 13 2310132 50 tuple
2 7201 10 403256 9 2713388 58 function
3 521 1 274376 6 2987764 64 dict (no owner)
4 197 0 249512 5 3237276 70 dict of module
5 3516 5 239088 5 3476364 75 types.CodeType
6 474 1 206060 4 3682424 79 type
7 474 1 189072 4 3871496 83 dict of type
8 7623 11 182952 4 4054448 87 __builtin__.cell
9 218 0 102608 2 4157056 89 dict of class
<269 more rows. Type e.g. '_.more' to view.>
>>> from django.utils.translation import ugettext_lazy
>>> tmp = [ugettext_lazy('foo%d' % i) for i in xrange(1000)]
>>> heapy.heap()
Partition of a set of 320759 objects. Total size = 16691776 bytes.
Index Count % Size % Cumulative % Kind (class / dict of class)
0 68201 21 3819256 23 3819256 23 function
1 129623 40 3110952 19 6930208 42 __builtin__.cell
2 78723 25 2595380 16 9525588 57 tuple
3 3521 1 2218760 13 11744348 70 dict (no owner)
4 1054 0 1762288 11 13506636 81 dict of django.utils.functional.__proxy__
5 26728 8 1726712 10 15233348 91 str
6 197 0 249512 1 15482860 93 dict of module
7 3516 1 239088 1 15721948 94 types.CodeType
8 474 0 206060 1 15928008 95 type
9 474 0 189072 1 16117080 97 dict of type
<269 more rows. Type e.g. '_.more' to view.>
