missing frame.f_lineno on JUMP_ABSOLUTE

[graingert]

BPO	47085
Nosy	@nedbat, @markshannon, @graingert

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2022-03-22.11:00:07.217>
created_at = <Date 2022-03-21.19:21:07.574>
labels = ['invalid', '3.10', '3.11']
title = 'missing frame.f_lineno on JUMP_ABSOLUTE'
updated_at = <Date 2022-03-22.11:00:07.216>
user = 'https://github.com/graingert'

bugs.python.org fields:

activity = <Date 2022-03-22.11:00:07.216>
actor = 'Mark.Shannon'
assignee = 'none'
closed = True
closed_date = <Date 2022-03-22.11:00:07.217>
closer = 'Mark.Shannon'
components = []
creation = <Date 2022-03-21.19:21:07.574>
creator = 'graingert'
dependencies = []
files = []
hgrepos = []
issue_num = 47085
keywords = []
message_count = 2.0
messages = ['415697', '415759']
nosy_count = 3.0
nosy_names = ['nedbat', 'Mark.Shannon', 'graingert']
pr_nums = []
priority = 'normal'
resolution = 'not a bug'
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue47085'
versions = ['Python 3.10', 'Python 3.11']

[graingert]

the following code prints:

import sys
import dis

import pprint

def demo():
    for i in range(1):
        if i >= 0:
            pass


class Tracer:
    def __init__(self):
        self.events = []

    def trace(self, frame, event, arg):
        self.events.append((frame.f_lineno, frame.f_lasti, event))
        frame.f_trace_lines = True
        frame.f_trace_opcodes = True
        return self.trace

def main():
    t = Tracer()
    old_trace = sys.gettrace()
    try:
        sys.settrace(t.trace)
        demo()
    finally:
        sys.settrace(old_trace)
    dis.dis(demo)
    pprint.pp(t.events)


if __name__ == "__main__":
    sys.exit(main())

7 0 LOAD_GLOBAL 0 (range)
2 LOAD_CONST 1 (1)
4 CALL_FUNCTION 1
6 GET_ITER
>> 8 FOR_ITER 7 (to 24)
10 STORE_FAST 0 (i)

8 12 LOAD_FAST 0 (i)
14 LOAD_CONST 2 (0)
16 COMPARE_OP 5 (>=)
18 POP_JUMP_IF_FALSE 11 (to 22)

9 20 NOP
>> 22 JUMP_ABSOLUTE 4 (to 8)

7 >> 24 LOAD_CONST 0 (None)
26 RETURN_VALUE
[(6, -1, 'call'),
(7, 0, 'line'),
(7, 0, 'opcode'),
(7, 2, 'opcode'),
(7, 4, 'opcode'),
(7, 6, 'opcode'),
(7, 8, 'opcode'),
(7, 10, 'opcode'),
(8, 12, 'line'),
(8, 12, 'opcode'),
(8, 14, 'opcode'),
(8, 16, 'opcode'),
(8, 18, 'opcode'),
(9, 20, 'line'),
(9, 20, 'opcode'),
(None, 22, 'opcode'),
(7, 8, 'line'),
(7, 8, 'opcode'),
(7, 24, 'opcode'),
(7, 26, 'opcode'),
(7, 26, 'return')]

but I'd expect (9, 22, 'opcode') instead of (None, 22, 'opcode'),

[markshannon]

The JUMP_ABSOLUTE doesn't have a line number, as it doesn't correspond to any source.

The jump back to the top could follow either the `if i >= 0:` or the `pass`, so cannot have a line number.

Don't expect every bytecode to map directly back to the source, especially if it cannot raise an exception.

Per-line tracing is well defined by PEP-626. Per-opcode tracing is not.

[graingert]

looks like this was fixed by ddc4a78 #93229