I want to add a test by setting frame.f_lineno in a callback function with sys.monitoring.register_callback. But as the callback is invoked by the vector calling convention, the frame is owned by thread (not interpreter), which makes WITHIN_STACK_BOUNDS fail in setattr with --with-pydebug build.

We should probably raise an exception if we try to set the line number of the currently executing frame.

Yeah, and I try to raise a RuntimeError in frame.f_lineno setter. But the interpreter cannot even reach there as it fails at the STORE_ATTR bytecode. So maybe we should have a document suggesting users not to setattr on current frame (i.e., the callback frame itself) in event callback?

We should probably raise an exception if we try to set the line number of the currently executing frame.

Yeah, and I try to raise a RuntimeError in frame.f_lineno setter. But the interpreter cannot even reach there as it fails at the STORE_ATTR bytecode. So maybe we should have a document suggesting users not to setattr on current frame (i.e., the callback frame itself) in event callback?

frame_lineno_set_impl should not modify the stack if it raises, so the assertion should pass.

No need to do that in this PR though.

Thanks @xuantengh for the PR, and @markshannon for merging it 🌮🎉.. I'm working now to backport this PR to: 3.14.
🐍🍒⛏🤖 I'm not a witch! I'm not a witch!

GH-137280 is a backport of this pull request to the 3.14 branch.

Sorry for the late reply, I just found the lineno setter may modify the frame stack of the function being monitored, so the frame stack pointer may be altered after the callback, which makes the WITHIN_STACK_BOUNDS checks fail in some bytecodes. But as you said, we may need to do it in another PR.