A Better PROFiler
A Better PROFiler
- Free software: MIT license
- Documentation: https://bprof.readthedocs.io.
The reason bprof exists is that the major Python profiling packages simply don't profile robustly. They use timestamps and ad-hoc methods for keeping track of how time passes. For example, one approach is to timestamp when a function starts and when it stops, and then count this as the function time. This includes time spent in the profiling hooks.
The approach taken here is to integrate all of the time between hooks and add it to the appropriate records. By registering for all hooks except for opcodes, the time spent out of the profiler is directly measured. The time is measured right after entering bprof, and right before exiting. This allows for as rigorous time measurement as possible. Then, stacks are used to track the current contexts and record detailed profiling information.
import bprof._bprof as bp import time def f(): time.sleep(1) bp.start() f() bp.stop() bp.dump("")
Name: f, 1.2142e-05 1.00074(6.21e-07/1.00074): time.sleep(1) Name: <built-in function stop>, 0 Name: <built-in function sleep>, 1.00074
There is a lot of future work. This is just a first pass.
- Save statistics
- First release on PyPI.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size bprof-0.5.2-cp37-cp37m-macosx_10_14_x86_64.whl (24.4 kB)||File type Wheel||Python version cp37||Upload date||Hashes View|
|Filename, size bprof-0.5.2.tar.gz (13.7 kB)||File type Source||Python version None||Upload date||Hashes View|
Hashes for bprof-0.5.2-cp37-cp37m-macosx_10_14_x86_64.whl