Repository navigation
bpo-33165: Remove redundant stack unwind for findCaller() - #17714
evandrocoan wants to merge 2 commits into
Conversation
Sorry, something went wrong.
the right frame on Lib/logging/__init__.py:currentframe()
because when extending the default Logger and implementing/specializing the `_log` function, all stacktraces need to be increased by 1. For example, the debug_tools pypi module inherits from Logger and defines its own _log() function: https://github.com/evandrocoan/debugtools/blob/d279bf3278f501294a72159f3aa189b7237528b2/all/debug_tools/logger.py#L166 https://github.com/evandrocoan/debugtools/blob/d279bf3278f501294a72159f3aa189b7237528b2/all/debug_tools/logger.py#L970 https://github.com/evandrocoan/debugtools/blob/d279bf3278f501294a72159f3aa189b7237528b2/all/debug_tools/logger.py#L1317
bpo-33165: Remove redundant stack unwind for findCaller() python/cpython#17714
bpo-33165: Remove redundant stack unwind for findCaller() python/cpython#17714
| file name, line number and function name. | ||
| """ | ||
| f = currentframe() | ||
| f = currentframe(self.default_stack_level + stacklevel) |
There was a problem hiding this comment.
Shouldn't the unwinding really be done later? A few lines down, we read:
while hasattr(f, "f_code"):
co = f.f_code
filename = os.path.normcase(co.co_filename)
if filename == _srcfile:
f = f.f_back
continueThis 'escapes' the stack frames generated inside the current file. I would expect any additional stacklevel treatment to happen afterwards.
As a result of the current implementation (with or without the proposed changes in the current PR), the stacklevel parameter to .warn must be one higher than the parameter to .warning in order to get to the same frame.
Sorry, something went wrong.
| raise Exception | ||
| except Exception: | ||
| return sys.exc_info()[2].tb_frame.f_back | ||
| return sys.exc_info()[level-1].tb_frame.f_back |
There was a problem hiding this comment.
I believe this is incorrect. exc_info returns three items and probably all we can do is
| return sys.exc_info()[level-1].tb_frame.f_back | |
| frame = sys.exc_info()[2].tb_frame.f_back | |
| while frame and stacklevel > 0: | |
| frame = frame.f_back | |
| stacklevel -= 1 | |
| if not frame: | |
| raise ValueError("call stack is not deep enough") | |
| return frame |
Sorry, something went wrong.
|
This was already fixed/implemented by dde9fdb |
Sorry, something went wrong.
by directly getting the right frame on Lib/logging/init.py:currentframe()
This is a slight improvement for #7424 (bpo-33165: Added stacklevel parameter to logging APIs)
Instead of getting the fullstack trace and only then unwind the desired frames, just pass the required frame index and get it directly, i.e., without "unstacking" n frames.
I also added the frame level as a Logger attribute because when extending the default Logger and implementing/specializing the _log function, all stacktraces need to be increased by 1. For example, the debug_tools pypi module inherits from Logger and defines its own _log() function:
https://github.com/evandrocoan/debugtools/blob/d279bf3278f501294a72159f3aa189b7237528b2/all/debug_tools/logger.py#L166
https://github.com/evandrocoan/debugtools/blob/d279bf3278f501294a72159f3aa189b7237528b2/all/debug_tools/logger.py#L970
https://github.com/evandrocoan/debugtools/blob/d279bf3278f501294a72159f3aa189b7237528b2/all/debug_tools/logger.py#L1317
https://bugs.python.org/issue33165