| Age | Commit message (Collapse) | Author |
|
Detect when `SimpleLazyObject._setupfunc` is a bound method of the same
instance to use a safe representation and avoid infinite recursion.
|
|
|
|
This allows for formatting of lazy objects which have a custom formatter
defined by overriding the default implementation from `object`.
|
|
|
|
We can avoid the function call and assignment overhead which could be
significant when instantiating many lazy objects. It's still easy enough
to read too.
|
|
Previously, the proxy class was prepared lazily:
lazy_identity = lazy(identity, int)
lazy_identity(10) # prepared here
lazy_identity(10)
This has a slight advantage that if the lazy doesn't end up getting
used, the preparation work is skipped, however that's not very likely.
Besides this laziness, it is also inconsistent in that the methods which
are wrapped directly (__str__ etc.) are prepared already when __proxy__
is defined, and there is a weird half-initialized state.
This change it so that everything is prepared already on the first line
of the example above.
|
|
This order reads more naturally and puts methods into three groups:
1. The methods required to support the implementation of __proxy__, e.g.
__deepcopy__ doesn't come from `object` and __reduce__ is
overridden to support behavior required explicitly for pickling of
lazy objects.
2. Methods that are specifically overridden from `object` which we
don't want to inherit from the provided resultclasses. These will be
skipped later when we add methods from the resultclasses.
3. Additional methods - that is _add__, __radd__, and __mod__ - don't
come from `object`, but typically from `str` and `int` which are the
most common use cases.
Co-authored-by: Nick Pope <nick@nickpope.me.uk>
|
|
@total_ordering is slow. Using the following micro-benchmark
(resultclasses intentionally omitted to narrow the scope):
import cProfile
from django.utils.functional import lazy
def identity(x): return x
cProfile.run("for i in range(10000): str(lazy(identity)(1))")
Before:
380003 function calls in 0.304 seconds
ncalls tottime percall cumtime percall filename:lineno(function)
1 0.016 0.016 0.304 0.304 <string>:1(<module>)
10000 0.002 0.000 0.002 0.000 bench.py:5(double)
10000 0.005 0.000 0.006 0.000 functional.py:100(__cast)
10000 0.007 0.000 0.013 0.000 functional.py:106(__str__)
10000 0.005 0.000 0.017 0.000 functional.py:140(__wrapper__)
10000 0.020 0.000 0.258 0.000 functional.py:60(lazy)
10000 0.039 0.000 0.039 0.000 functional.py:68(__proxy__)
10000 0.010 0.000 0.012 0.000 functional.py:77(__init__)
10000 0.002 0.000 0.002 0.000 functional.py:84(__prepare_class__)
10000 0.025 0.000 0.075 0.000 functools.py:186(total_ordering)
10000 0.015 0.000 0.028 0.000 functools.py:189(<setcomp>)
10000 0.024 0.000 0.044 0.000 functools.py:37(update_wrapper)
10000 0.005 0.000 0.005 0.000 functools.py:67(wraps)
10000 0.074 0.000 0.114 0.000 {built-in method builtins.__build_class__}
1 0.000 0.000 0.304 0.304 {built-in method builtins.exec}
150000 0.023 0.000 0.023 0.000 {built-in method builtins.getattr}
10000 0.004 0.000 0.004 0.000 {built-in method builtins.max}
80000 0.025 0.000 0.025 0.000 {built-in method builtins.setattr}
1 0.000 0.000 0.000 0.000 {method 'disable' of '_lsprof.Profiler' objects}
10000 0.003 0.000 0.003 0.000 {method 'update' of 'dict' objects}
After:
240003 function calls in 0.231 seconds
ncalls tottime percall cumtime percall filename:lineno(function)
1 0.016 0.016 0.231 0.231 <string>:1(<module>)
10000 0.002 0.000 0.002 0.000 bench.py:5(double)
10000 0.006 0.000 0.012 0.000 functional.py:105(__str__)
10000 0.005 0.000 0.017 0.000 functional.py:159(__wrapper__)
10000 0.015 0.000 0.186 0.000 functional.py:60(lazy)
10000 0.022 0.000 0.022 0.000 functional.py:68(__proxy__)
10000 0.010 0.000 0.012 0.000 functional.py:76(__init__)
10000 0.002 0.000 0.002 0.000 functional.py:83(__prepare_class__)
10000 0.004 0.000 0.006 0.000 functional.py:99(__cast)
10000 0.023 0.000 0.043 0.000 functools.py:37(update_wrapper)
10000 0.004 0.000 0.004 0.000 functools.py:67(wraps)
10000 0.102 0.000 0.124 0.000 {built-in method builtins.__build_class__}
1 0.000 0.000 0.231 0.231 {built-in method builtins.exec}
70000 0.011 0.000 0.011 0.000 {built-in method builtins.getattr}
50000 0.007 0.000 0.007 0.000 {built-in method builtins.setattr}
1 0.000 0.000 0.000 0.000 {method 'disable' of '_lsprof.Profiler' objects}
10000 0.003 0.000 0.003 0.000 {method 'update' of 'dict' objects}
|
|
They are no longer special cased.
|
|
|
|
Unnecessary since c716fe87821df00f9f03ecc761c914d1682591a2 and
7b2f2e74adb36a4334e83130f6abc2f79d395235.
|
|
may be bytes.
If the result type is bytes, then calling bytes() on it does nothing.
If the result type is not bytes, we should not cast to bytes, just
because the return value may be bytes.
|
|
This removes __text_cast() as it's the same as __cast().
_delegate_bytes and __delegate_text are mutually exclusive so the
`if self._delegate_bytes` branch in __cast() is unreachable.
Co-Authored-By: David Sanders <shang.xiao.sanders@gmail.com>
|
|
Unused since c716fe87821df00f9f03ecc761c914d1682591a2.
|
|
django.utils.functional.cached_property().
Per deprecation timeline.
|
|
This avoids an extra __getattribute__() call for self._wrapped.
|
|
Regression in 97d7990abde3fe4b525ae83958fd0b52d6a1d13f.
Co-authored-by: Mariusz Felisiak <felisiak.mariusz@gmail.com>
Co-authored-by: Theo Alexiou <theofilosalexiou@gmail.com>
|
|
Thanks Sergey Fedoseev for the initial patch.
|
|
|
|
|
|
|
|
|
|
Co-authored-by: Claude Paroz <claude@2xlibre.net>
|
|
|
|
|
|
lazy() should prepare the proxy class only once (the first time it's
used) not on every call.
Regression in b4e76f30d12bfa8a53cc297c60055c6f4629cc4c.
|
|
|
|
|
|
Co-Authored-By: Sergey Fedoseev <fedoseev.sergey@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
iter(dict) is equivalent to iter(dict.keys()).
|
|
|
|
|
|
Thanks Tim Graham for the review.
|
|
|
|
|
|
|
|
|
|
__ne__() defaults to the opposite of __eq__() on Python 3
when it doesn't return NotImplemented.
|
|
Thanks Tim Graham for the review.
|
|
Thanks Tim Graham and Simon Charette for the reviews.
|
|
Thanks Tim Graham for the review.
|
|
|
|
|
|
|