Table of Contents
速 means fast in Chinese.
su-uuid is an uuid CPython Extension implemented in Rust, which is compatable with the CPython uuid module.
By compatability, I mean this module is using the CPython test cases for uuid module.
You can consider this as a drop-in replacement.
Incompatabe part:
- CPython 3.7 add
UUID.is_safeattribute, we don't have it - some mismatch arguments are raise different exception, CPython raise ValueError, we raise TypeError, e.g. passing
strtobytesargument
As a personal practice for using Rust, PyO3 to write CPython extension. And the reason to choose uuid module is that this module is quite simple, but also contains mutiple things we will need to handle. The simple part is that the module does not contains a lot of stuff, just few functions and a class.
The good part for practicing is that it has:
- a new class and multiple methods, getter
- multiple functions
- module level constant object
- default arguments, positional arguments, keyword arguments
- good test cases in CPython
- Rust has a
uuidcrate
As this module can pass the CPython test cases for uuid module, this is already usable for any use case that you want to use uuid .
import su_uuid as uuid
uuid.uuid4()Exceptions:
>>> import su_uuid as u
>>> u.UUID('000z0000000000000000000000000000')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: UUID creation has error: Invalid character; found `z` (0x7a) at offset 3rustup install nightly-2018-07-16
rustup override set nightly-2018-07-16
make wheelmake installmake testmake benchThere are some micro benchmark in the benchmark/ folder. They are written with pytest-benchmark, you can see the benchmark result by running make bench or the pytest benchmark/.
Here is the rough speedup compare to the pure Python implementation in CPython:
----------------------------------------------------------------------------------------- benchmark 'uuid1': 2 tests -----------------------------------------------------------------------------------------
Name (time in ns) Min Max Mean StdDev Median IQR Outliers OPS (Kops/s) Rounds Iterations
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_uuid1_rs 596.0001 (1.0) 18,153.9999 (1.0) 697.1010 (1.0) 277.6190 (1.0) 692.0000 (1.0) 29.9997 (1.0) 6;567 1,434.5123 (1.0) 7521 1
test_uuid1_py 8,145.9993 (13.67) 328,017.9999 (18.07) 9,133.8569 (13.10) 7,114.5530 (25.63) 8,656.0003 (12.51) 439.7500 (14.66) 70;953 109.4828 (0.08) 12211 1
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
----------------------------------------------------------------------------------------- benchmark 'uuid3': 2 tests ----------------------------------------------------------------------------------------
Name (time in ns) Min Max Mean StdDev Median IQR Outliers OPS (Kops/s) Rounds Iterations
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_uuid3_rs 679.0006 (1.0) 348,720.9997 (9.67) 811.5521 (1.0) 1,684.4731 (1.57) 776.0000 (1.0) 28.0006 (1.0) 123;7777 1,232.2067 (1.0) 168436 1
test_uuid3_py 4,395.9999 (6.47) 36,069.0001 (1.0) 4,775.2736 (5.88) 1,072.4803 (1.0) 4,606.0004 (5.94) 99.0003 (3.54) 1013;1962 209.4121 (0.17) 27263 1
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------- benchmark 'uuid4': 2 tests ---------------------------------------------------------------------------------------
Name (time in ns) Min Max Mean StdDev Median IQR Outliers OPS (Kops/s) Rounds Iterations
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_uuid4_rs 246.0001 (1.0) 15,007.0000 (1.0) 294.9147 (1.0) 350.7388 (1.0) 281.0002 (1.0) 8.0008 (1.0) 171;5594 3,390.8115 (1.0) 21173 1
test_uuid4_py 2,748.0000 (11.17) 26,522.9992 (1.77) 2,997.0379 (10.16) 617.0421 (1.76) 2,936.0008 (10.45) 68.0011 (8.50) 1188;4438 333.6628 (0.10) 66437 1
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------- benchmark 'uuid5': 2 tests ---------------------------------------------------------------------------------------
Name (time in ns) Min Max Mean StdDev Median IQR Outliers OPS (Kops/s) Rounds Iterations
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_uuid5_rs 664.9998 (1.0) 24,268.9994 (1.0) 768.2647 (1.0) 262.3117 (1.0) 752.9998 (1.0) 26.0006 (1.0) 1690;6876 1,301.6347 (1.0) 176274 1
test_uuid5_py 4,345.0000 (6.53) 46,773.0006 (1.93) 4,654.5543 (6.06) 784.7173 (2.99) 4,577.0003 (6.08) 93.9990 (3.62) 479;2298 214.8433 (0.17) 38049 1Apache 2.0