Bug report
Every recursive call of Python function increases the recursion_depth counter. Some recursive calls of C functions also increase it. A RecursionError is raised when it exceed the limit (sys.getrecursionlimit()).
The following code expectedly prints a value close to 1/2 of sys.getrecursionlimit() in all Python versions before 3.7 (including 2.7):
def test(n):
if n > 0:
a = []
a.extend(test(n-1) for _ in [1])
for i in range(1000):
try:
test(i)
except RecursionError:
print(i)
breakOne for every test() function call and one for every generator expression. Both are shown in a traceback.
In Python 3.7 to 3.11 it prints 1/3 of the limit. Perhaps it adds one level for every list.extend() call?
In Python 3.12 it returns to 1/2 of the limit.
But in the main branch it only prints 1/6 of the limit. How can it increase the recursion_depth counter 6 times per the recursion level? It is a regression, because now the code that was successfully passed in previous versions starts to raise a RecursionError. See #111803 for example.
Bug report
Every recursive call of Python function increases the
recursion_depthcounter. Some recursive calls of C functions also increase it. A RecursionError is raised when it exceed the limit (sys.getrecursionlimit()).The following code expectedly prints a value close to 1/2 of
sys.getrecursionlimit()in all Python versions before 3.7 (including 2.7):One for every
test()function call and one for every generator expression. Both are shown in a traceback.In Python 3.7 to 3.11 it prints 1/3 of the limit. Perhaps it adds one level for every
list.extend()call?In Python 3.12 it returns to 1/2 of the limit.
But in the main branch it only prints 1/6 of the limit. How can it increase the
recursion_depthcounter 6 times per the recursion level? It is a regression, because now the code that was successfully passed in previous versions starts to raise a RecursionError. See #111803 for example.