Отладка в Python 3.6: Быстрее, Выше, Сильнее

В данном докладе мы узнаем, как работает новый интерфейс для вычисления фреймов, как он может помочь при создании быстрого отладчика, и почему такой быстрый отладчик невозможно было создать в предыдущих версиях языка Python. Для тех же, кто ещё не принял окончательное решение о переходе на Python 3.6, этот доклад даст несколько дополнительных причин, почему это стоит сделать.

Transcript

1. Debugging in Python 3.6: Better, Faster, Stronger Elizaveta Shashkova JetBrains

2. Bio • Software developer of PyCharm IDE at JetBrains • Debugger • Saint Petersburg 2
3. Debugging • Adding print statements • Logging 3
4. Debugging • Adding print statements • Logging • Special tools: debuggers 4
5. Debugger’s Performance 5 Run Debug Debuggers are 30 times slower
6. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 6
7. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 7
8. 8 def tracefunc(frame, event, arg): print(frame.f_lineno, event) return tracefunc sys.settrace(tracefunc) 1 2 3 4 5 6 sys.settrace(tracefunc) - set the system tracing function Tracing Function
9. def foo(): friends = ["Bob", "Tom"] for f in friends: print("Hi %s!” % f) return len(friends) sys.settrace(tracefunc) foo() 1 2 3 4 5 6 7 8 9 Tracing Function 9
10. 1 2 3 4 5 6 7 8 9 1 call def foo(): friends = ["Bob", "Tom"] for f in friends: print("Hi %s!” % f) return len(friends) sys.settrace(tracefunc) foo() Tracing Function 10
11. 1 2 3 4 5 6 7 8 9 1 call 2 line def foo(): friends = ["Bob", "Tom"] for f in friends: print("Hi %s!” % f) return len(friends) sys.settrace(tracefunc) foo() Tracing Function 11
12. 1 2 3 4 5 6 7 8 9 1 call 2 line 3 line 4 line Hi Bob! def foo(): friends = ["Bob", "Tom"] for f in friends: print("Hi %s!” % f) return len(friends) sys.settrace(tracefunc) foo() Tracing Function 12
13. 1 2 3 4 5 6 7 8 9 1 call 2 line 3 line 4 line Hi Bob! 3 line 4 line Hi Tom! def foo(): friends = ["Bob", "Tom"] for f in friends: print("Hi %s!” % f) return len(friends) sys.settrace(tracefunc) foo() Tracing Function 13
14. 1 2 3 4 5 6 7 8 9 1 call 2 line 3 line 4 line Hi Bob! 3 line 4 line Hi Tom! 5 line 5 return def foo(): friends = ["Bob", "Tom"] for f in friends: print("Hi %s!” % f) return len(friends) sys.settrace(tracefunc) foo() Tracing Function 14
15. Build Python Debugger • Breakpoints • Stepping 15
16. Tracing Debugger • Suspend program if breakpoint’s line equals frame.f_lineno • Handle events for stepping 16
17. Performance def foo(): friends = ["Bob", "Tom"] for f in friends: print("Hi %s!” % f) return len(friends) sys.settrace(tracefunc) foo() 1 2 3 4 5 6 7 8 9 1 call 2 line 3 line 4 line Hi Bob! 3 line 4 line Hi Tom! 5 line 5 return 17
18. Example 1 def calculate(): sum = 0 for i in range(10 ** 7): sum += i return sum 1 2 3 4 5 6 7 8 9 18
19. Example 1 def calculate(): sum = 0 for i in range(10 ** 7): sum += i return sum def tracefunc(frame, event, arg): return tracefunc 1 2 3 4 5 6 7 8 9 19
20. Performance 20 Run Tracing Breakpoints 0,80 sec 6,85 sec 19,81 sec
21. Performance 21 Run Tracing Breakpoints 0,80 sec 6,85 sec 19,81 sec
22. Performance 22 Run Tracing Breakpoints 0,80 sec 6,85 sec 19,81 sec
23. Performance 23 Run Tracing Breakpoints 0,80 sec 6,85 sec 19,81 sec ~ 25 times slower!
24. Problem • Tracing call on every line 24
25. Optimizations 25
26. Optimization 1 • Skip scopes without breakpoints • frame.f_code.co_name - name of the function 26
27. Optimization 2 • Rewrite tracing function in C 27
28. Optimization 2 • Rewrite tracing function in C • Works only for CPython • ~ 4 times faster 28
29. Problem • Tracing call on every line 29
30. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 30
31. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 31
32. Python 3.6 32
33. Python 3.6 • New frame evaluation API • PEP 523 33
34. PEP 523 • Handle evaluation of frames • Add a new field to code objects • C API 34
35. Frame Evaluation cdef frame_eval(PyFrameObject *frame, int exc): func_name = <object> frame.f_code.co_name line_number = <object> frame.f_lineno print(line_number, func_name) return _PyEval_EvalFrameDefault(frame, exc) 1 2 3 4 5 6 7 8 9 35
36. cdef frame_eval(PyFrameObject *frame, int exc): func_name = <object> frame.f_code.co_name f line_number = <object> frame.f_lineno print(line_number, func_name) return _PyEval_EvalFrameDefault(frame, exc) 1 2 3 4 5 6 7 8 9 Frame Evaluation 36
37. cdef frame_eval(PyFrameObject *frame, int exc): func_name = <object> frame.f_code.co_name line_number = <object> frame.f_lineno print(line_number, func_name) return _PyEval_EvalFrameDefault(frame, exc) 1 2 3 4 5 6 7 8 9 Frame Evaluation 37
38. cdef frame_eval(PyFrameObject *frame, int exc): func_name = <object> frame.f_code.co_name line_number = <object> frame.f_lineno print(line_number, func_name) return _PyEval_EvalFrameDefault(frame, exc) 1 2 3 4 5 6 7 8 9 Frame Evaluation 38
39. cdef frame_eval(PyFrameObject *frame, int exc): func_name = <object> frame.f_code.co_name line_number = <object> frame.f_lineno print(line_number, func_name) return _PyEval_EvalFrameDefault(frame, exc) 1 2 3 4 5 6 7 8 9 Frame Evaluation 39
40. cdef frame_eval(PyFrameObject *frame, int exc): func_name = <object> frame.f_code.co_name line_number = <object> frame.f_lineno print(line_number, func_name) return _PyEval_EvalFrameDefault(frame, exc) def set_frame_eval(): cdef PyThreadState *state = \ PyThreadState_Get() state.interp.eval_frame = frame_eval 1 2 3 4 5 6 7 8 9 Frame Evaluation 40
41. Example 1 2 3 4 5 6 7 8 9 10 11 def first(): second() def second(): third() def third(): pass set_frame_eval() first() 41
42. Example def first(): second() def second(): third() def third(): pass set_frame_eval() first() 1 2 3 4 5 6 7 8 9 10 11 1 first 4 second 7 third 42
43. Custom Frame Evaluation • It works • Executed while entering a frame • Access to frame and code object 43
44. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 44
45. Problem • Tracing call on every line 45
46. Problem • Tracing call on every line • Remove the tracing function 46
47. Replace tracing function with frame evaluation function 47
48. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 48
49. Build Python Debugger • Breakpoints • Stepping 49
50. Breakpoints • Access to the whole code object 50
51. Breakpoints • Access to the whole code object • Insert breakpoint’s code into frame’s code 51
52. Breakpoints def maximum(a, b): if a > b: return a else: return b 1 2 3 4 5 6 7 8 52
53. Breakpoints def maximum(a, b): if a > b: return a # breakpoint else: return b 1 2 3 4 5 6 7 8 53
54. Breakpoints def maximum(a, b): if a > b: return a # breakpoint else: return b 1 2 3 4 5 6 7 8 breakpoint() 54
55. Breakpoints def maximum(a, b): if a > b: breakpoint() return a # breakpoint else: return b 1 2 3 4 5 6 7 8 55
56. Python Bytecode def maximum(a, b): if a > b: return a else: return b import dis dis.dis(maximum) 1 2 3 4 5 6 7 8 56
57. Python Bytecode def maximum(a, b): if a > b: return a else: return b import dis dis.dis(maximum) 1 2 3 4 5 6 7 8 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE 57
58. Python Bytecode def maximum(a, b): if a > b: return a else: return b import dis dis.dis(maximum) 1 2 3 4 5 6 7 8 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE 58
59. Python Bytecode def maximum(a, b): if a > b: return a else: return b import dis dis.dis(maximum) 1 2 3 4 5 6 7 8 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE 59
60. Python Bytecode def maximum(a, b): if a > b: return a else: return b import dis dis.dis(maximum) 1 2 3 4 5 6 7 8 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE 60
61. Python Bytecode def maximum(a, b): if a > b: return a else: return b import dis dis.dis(maximum) 1 2 3 4 5 6 7 8 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE 61
62. Python Bytecode 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE 62
63. 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE Python Bytecode 63
64. 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE Python Bytecode 64
65. 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE Python Bytecode 65
66. Bytecode Modification 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE breakpoint() 66
67. Bytecode Modification • Insert breakpoint’s code • Update arguments and offsets 67
68. Bytecode Modification • Insert breakpoint’s code • Update arguments and offsets • 200 lines in Python 68
69. Bytecode Modification 2 0 LOAD_FAST 0 (a) 2 LOAD_FAST 1 (b) 4 COMPARE_OP 4 (>) 6 POP_JUMP_IF_FALSE 12 3 8 LOAD_FAST 0 (a) 10 RETURN_VALUE 5 >> 12 LOAD_FAST 1 (b) 14 RETURN_VALUE 16 LOAD_CONST 0 (None) 18 RETURN_VALUE breakpoint() ?! 69
70. Breakpoint Bytecode def _stop_at_break(): # a lot of code here def breakpoint(): _stop_at_break() 1 2 3 4 5 6 7 8 0 LOAD_GLOBAL 0 2 CALL_FUNCTION 0 4 POP_TOP 6 LOAD_CONST 0 8 RETURN_VALUE 70
71. Build Python Debugger • Breakpoints • Stepping 71
72. Stepping • Inserting temporary breakpoint on every line • Use old tracing function 72
73. Frame evaluation debugger is ready 73
74. Example 1 def calculate(): sum = 0 for i in range(10 ** 7): sum += i return sum 1 2 3 4 5 74
75. Example 1 75 Run Tracing Frame 
 evaluation 0,80 sec 19,81 sec 0,81 sec
76. Example 1 76 Run Tracing Frame 
 evaluation 0,80 sec 19,81 sec 0,81 sec
77. Example 2 def foo(): pass def calculate(): sum = 0 for i in range(10 ** 7): foo() sum += i return sum 1 2 3 4 5 6 7 8 9 77
78. Example 2 78 Run Tracing Frame 
 evaluation 1,73 sec 43,58 sec 37,41 sec
79. PEP 523 • Handle evaluation of frames • Add a new field to code objects 79
80. PEP 523 • Expand PyCodeObject struct • co_extra - “scratch space” for the code object • Mark frames without breakpoints 80
81. Mark Frames Cython 1 2 3 4 5 6 7 8 9 81 cdef frame_eval(PyFrameObject *frame, int exc): flag = _PyCode_GetExtra(frame.f_code, index) if flag == NO_BREAKS_IN_FRAME: return _PyEval_EvalFrameDefault(frame, exc) # check for breakpoints ...
82. Example 2 82 Run Tracing Frame 
 evaluation 1,73 sec 43,58 sec 1,91 sec
83. PEP 523 • Handle evaluation of frames • Add a new field to code objects 83
84. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 84
85. Contents • Tracing debugger • Python 3.6 • Frame evaluation debugger • Results 85
86. Real Life Example 86
87. Real Life Example • Included into PyCharm 2017.1 • Works in production 87
88. PyCharm 88 Tracing
 w/o Cython Tracing
 with Cython Frame 
 evaluation 11,59 sec 5,66 sec 0,28 sec
89. Disadvantages • More complicated • Only with CPython • Only with Python 3.6 89
90. Frame Evaluation • Let’s move to Python 3.6! 90
91. Frame Evaluation • Let’s move to Python 3.6! • Let’s find another use cases 91
92. Use cases def maximum(a, b): if a > b: return a # breakpoint else: return b 1 2 3 4 5 6 7 8 breakpoint() 92
93. PEP 523 • Pyjion project • JIT for Python 93
94. Frame Evaluation • Let’s move to Python 3.6! • Let’s find another use cases 94
95. Links • Prototype: https://github.com/ Elizaveta239/frame-eval • PyCharm Community Edition source code • bytesinsert on PyPi 95
96. • Prototype: https://github.com/ Elizaveta239/frame-eval • PyCharm Community Edition source code • bytesinsert on PyPi Questions? 96 @lisa_shashkova