Python has dynamic scoping.

I recognize I should probably have learned this in my career by now. But chalk one more point up on my “Python is braindead” list.

def step(message):
    print(message)

class Step(object):
    def run(self, context):
        print(self == step)
        print(self)
        print(step)
        step("1")

if __name__ == "__main__":
    context = {}
    procedure = [
        Step(),
    ]
    for step in procedure:
        step.run(context)
    print("Done.")

Running this with Python 3.10.6 produces the following output:

True
<__main__.Step object at 0x000001D65110BFD0>
<__main__.Step object at 0x000001D65110BFD0>
Traceback (most recent call last):
  File "E:\Source\marisalactation\python-does-not-have-lexical-scoping.py", line 17, in <module>
    step.run(context)
  File "E:\Source\marisalactation\python-does-not-have-lexical-scoping.py", line 9, in run
    step("1")
TypeError: 'Step' object is not callable

In other words, step("1") does not call the function defined a few lines above but rather at that point in code step references the for loop iteration variable step defined at the bottom of the file. Congratulations Python is, at least here, dynamically scoped.

I forget that languages do that.

The issue is, of course, not the presence of dynamic scoping (though I prefer visual signifiers to such) but the odd interaction between various scoping rules.

EDIT:

I guess likely what is happening here is that def step binds that function implementation to the step global variable and then for step inside the if block is rebinding step with its iteration variable. Rather than actually having dynamic scoping.