Before PEP 669, the opcode trace is togglable after trace function being set. So something like

importsysdefopcode_trace_func(frame,event,arg):print(frame,event)returnopcode_trace_funcsys.settrace(opcode_trace_func)sys._getframe().f_trace=opcode_trace_funcsys._getframe().f_trace_opcodes=Truedeff():a= [1,2,3]returna[0]f()

would work.

After PEP 669, the opcode event is set onsys.settrace() so it's impossible to enable it after enabling the trace.

sys._getframe().f_trace_opcodes=Truesys.settrace(opcode_trace_func)sys._getframe().f_trace=opcode_trace_func

kind of works but it's super intuitive.

For the current implementation, if we enabled opcode before we dosys.settrace(), it would be "enabled" forever. Meaning the event will be always on, the callback will be always triggered, butframe.f_trace_opcode will be checked in the callback function. Can we always have opcode callback on? It would be a performance hit(extra call for each opcode), but that happens if we want any opcode event at all, and I think that's basically how the old tracing system does it.

Another option would be makingf_trace_opcode a descriptor and toggle(actually, probable just enable) opcode events there.

If we don't want to do that at all and just want to break the backward-compatibility because no one is doing insturction trace anyway (I only realized this when syncing#103050 to the lastest change), we should probably document it - make sure the current frame'sf_trace_opcode is set before callingsys.settrace() if you want opcode events.

BTW we would've caught this if we already had the instruction level debugging :)

Linked PRs

• gh-103615: Use local events for opcode tracing #109472