03.04.20 18:13, joannah nanjekye пише:
From my CS theory, a control flow graph models a program flow and one of its main characteristics is it has one entry and exit point. IIRC, CPython’s compilation process involves generation of a control flow graph.
Contrary to peephole optimizations, optimizations on the control flow graph are more global allowing us to have complex and global optimizations like branch and checkpoint eliminations etc.
I have seen several implementations of control flow optimizations. The one I am familiar with is the V8 control flow optimizer.
I tried to investigate this for one of my directed courses last fall but I want to know if there are people who have been thinking about this for CPython and what their thoughts are.
I implemented the CFG optimizer, but the problem was that it removed bytecode which served a marker for the lineno setter in the frame object. The lineno setter checked if it valid to jump to the specified instruction by counting instructions specific to some sybtax constructions. And if you have unconditional break, continue or return in the with or try block, the code for normal leaving the block was removed, and the lineno setter was not able to determine when we jump in or out of the block. Until we radically change the algorithm of the lineno setter it is hard to implement a true CFG optimizer. The peepholer contains special exceptions for such codes, but the CFG optimizer is more aggressive.