Add mechanism for getting active exception in a sys.monitoring PY_UNWIND callback (3.12)

[jpe]

I couldn't figure out a way get the active exception from a PY_UNWIND event callback -- I was hoping the exception would be passed to the callback but it isn't. This was an issue with the old tracer function debugger / profiler support so it's not really a regression, but I was hoping to avoid the work arounds that we used with it.

I mentioned this in #103082 but was encouraged to create a separate bug for it.

Linked PRs

• GH-106898: Add the exception as an argument to the PY_UNWIND event callback function. #107347
• [3.12] GH-106898: Add the exception as an argument to the PY_UNWIND event callback function. (GH-107347) #107382

[gaogaotiantian]

This should be doable by simply getting the exceptions from tstate. The question is whether to add it - it would change the interface of the monitoring callbacks and is not backward compatible (of course it won't break a lot of code as there aren't a lot yet). I'll leave the decision to @markshannon and if he's lazy to implement that, I can do it.

[jpe]

Before the callback is called, the current_exception is saved & set to NULL and then it is restored afterwards. It might be in one of the other tstate fields, but using sys.exc_info() doesn't retrieve it.

[gaogaotiantian]

Yes the current exception is saved. I meant we can do it otherwise in the instrumentation code to pass it down, it's not rocket science. This would be a design decision and Mark may have other concerns when he implemented this. That's why I said we should wait for his decision.

[markshannon]

It makes sense to pass the exception to the event handler.
We pass the exception for the RAISE event, so it would make sense to do so for PY_UNWIND.

Should C_RAISE event handlers also get the exception?

[jpe]

I think C_RAISE handlers should get the exception as an argument for consistency sake, if nothing else

[gaogaotiantian]

So we need to change the callback signatures. The feature is not even documented though (except for the PEP). That's probably a separate question but - we should have the official documentation for sys.monitoring right?

I can work on adding the exception to the callback if no one volunteers.